Stefan Müller - hu-berlin.de

Germanic syntaxStefan Müller

Textbooks in Language Sciences

Draft

of20thOctob

er2021, 18

:56

Textbooks in Language Sciences

Editors: Stefan Müller, Martin Haspelmath

Editorial Board: Claude Hagège, Marianne Mithun, Anatol Stefanowitsch, Foong Ha Yap

In this series:

1. Müller, Stefan. Grammatical theory: From transformational grammar to constraintbased

approaches.

2. Schäfer, Roland. Einführung in die grammatische Beschreibung des Deutschen.

3. Freitas, Maria João & Ana Lúcia Santos (eds.). Aquisição de língua materna e não

materna: Questões gerais e dados do português.

4. Roussarie, Laurent. Sémantique formelle: Introduction à la grammaire de Montague.

5. Kroeger, Paul. Analyzing meaning: An introduction to semantics and pragmatics.

6. Ferreira, Marcelo. Curso de semântica formal.

7. Stefanowitsch, Anatol. Corpus linguistics: A guide to the methodology.

ISSN: 23646209

Germanic syntaxStefan Müller

Müller, Stefan. 2021. Germanic syntax (Textbooks in Language Sciences). Berlin:Language Science Press.

This title can be downloaded at:http://langsci-press.org/catalog/book/000© 2021, Stefan MüllerPublished under the Creative Commons Attribution 4.0 Licence (CC BY 4.0):http://creativecommons.org/licenses/by/4.0/ISBN: no digital ISBN

no print ISBNs!ISSN: 2364-6209no DOIID not assigned!

Cover and concept of design: Ulrike HarbortTypesetting: Stefan MüllerFonts: Libertinus, Arimo, DejaVu Sans MonoTypesetting software: XƎLATEX

Language Science PressXhainGrünberger Str. 1610243 Berlin, Germanylangsci-press.org

Storage and cataloguing done by FU Berlin

http://langsci-press.org/catalog/book/000

http://langsci-press.org

For ???

Contents

Preface v

Acknowledgments xiii

Abbreviations xv

1 A general overview of the Germanic languages 11.1 Languages and speakers . . . . . . . . . . . . . . . . . . . . . . 11.2 Historical remarks and relatedness between the languages . . . 11.3 The three Germanic branches . . . . . . . . . . . . . . . . . . . 5

1.3.1 East Germanic . . . . . . . . . . . . . . . . . . . . . . . 51.3.2 North Germanic . . . . . . . . . . . . . . . . . . . . . . . 51.3.3 West Germanic . . . . . . . . . . . . . . . . . . . . . . . 7

2 Phenomena 112.1 Order of subject, object and verb . . . . . . . . . . . . . . . . . . 112.2 V2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.3 Scrambling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212.4 The position of adverbials . . . . . . . . . . . . . . . . . . . . . 222.5 Embedded clauses . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.5.1 Embedded clauses introduced by a complementizer . . . 232.5.2 Interrogative clauses . . . . . . . . . . . . . . . . . . . . 25

2.6 The use of expletives to mark the clause type . . . . . . . . . . . 272.7 Verbal Complexes in OV languages . . . . . . . . . . . . . . . . 282.8 Obligatoriness of subjects, case of subjects, and passives . . . . 292.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3 Phrase structure grammars and X theory 333.1 Symbols and rewrite rules . . . . . . . . . . . . . . . . . . . . . 333.2 Expanding PSG with features . . . . . . . . . . . . . . . . . . . . 403.3 Phrase structure rules for some aspects of German syntax . . . . 41

3.3.1 Noun phrases . . . . . . . . . . . . . . . . . . . . . . . . 42

Contents

3.3.2 Prepositional phrases . . . . . . . . . . . . . . . . . . . . 513.4 X theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4 Valence, argument order and adjunct placement 634.1 Valence representations . . . . . . . . . . . . . . . . . . . . . . . 634.2 Scrambling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.3 SVO: Languages with fixed SV order and valence features . . . . 694.4 Immediate dominance schemata . . . . . . . . . . . . . . . . . . 734.5 Scrambling and free VO/OV order . . . . . . . . . . . . . . . . . 784.6 Linear precedence rules . . . . . . . . . . . . . . . . . . . . . . . 784.7 Adjuncts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 814.8 Linking between syntax and semantics . . . . . . . . . . . . . . 884.9 Alternatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.9.1 CP/TP/VP models . . . . . . . . . . . . . . . . . . . . . . 89

5 The verbal complex 95

6 Verb position: Verb first and verb second 1076.1 The phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . 107

6.1.1 German as an SOV language . . . . . . . . . . . . . . . . 1096.1.2 Verb position in the Germanic SVO languages . . . . . . 1146.1.3 Verb second . . . . . . . . . . . . . . . . . . . . . . . . . 116

6.2 The analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1186.2.1 Verb first . . . . . . . . . . . . . . . . . . . . . . . . . . . 1186.2.2 Verb second . . . . . . . . . . . . . . . . . . . . . . . . . 121

7 Passive 1317.1 The phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . 131

7.1.1 Subjects and other subjects . . . . . . . . . . . . . . . . 1317.1.2 Comparison German, Danish, English, Icelandic . . . . . 137

7.2 The Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1427.2.1 Structural and lexical case and the Case Principle . . . . 1427.2.2 Argument reduction and case assignment: The passive . 1497.2.3 Argument extension and case assignment: AcI construc-

tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1507.2.4 Theoretical analysis of the crosslinguistic differences . . 152

7.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

ii Draft of 20th October 2021, 18:56

Contents

8 Clause types and expletives 1738.1 The phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

8.1.1 Embedded clauses introduced by a complementizer . . . 1738.1.2 Interrogative clauses . . . . . . . . . . . . . . . . . . . . 1758.1.3 The use of expletives to mark the clause type . . . . . . 177

8.2 The analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1798.2.1 Embedded clauses introduced by a complementizer . . . 1798.2.2 Interrogative clauses . . . . . . . . . . . . . . . . . . . . 1818.2.3 A lexical rule for the introduction of expletives . . . . . 184

8.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

9 Outlook 191

Appendix A: Solutions 193A.1 Phrase structure grammars and X theory . . . . . . . . . . . . . 193A.2 Valency, argument order and adjunct placement . . . . . . . . . 197A.3 The verbal complex . . . . . . . . . . . . . . . . . . . . . . . . . 204A.4 Verb position: Verb first and verb second . . . . . . . . . . . . . 207A.5 Passive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215A.6 Clause types and expletives . . . . . . . . . . . . . . . . . . . . . 215

References 217

Index 235Name index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235Language index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238

Draft of 20th October 2021, 18:56 iii

Preface

This book has two purposes: firstly the comparative analysis of the syntacticproperties of the Germanic languages and secondly the introduction of a spe-cific format for the description and comparison of languages. The framework inwhich the analyses are couched is called HPSG light. It is based on Head-DrivenPhrase Structure Grammar (HPSG) (Pollard & Sag 1987; 1994) in the specific ver-sion that is described in detail in Müller (2013b). However HPSG light does notcontain any complicated attribute value matrices (AVMs). If AVMs are used atall, they are reduced to the minimum containing a reduced set of features likeARG-ST for argument structure, COMPS for complements and SPR for specifier. Allother aspects of the analyses are represented in syntactic trees, which are easierto read. The idea behind the introduction of HPSG light is to provide some toolfor linguists who want to provide a more detailed description of a phenomenonwithout necessarily being forced to deal with all the technicalities. The degreeof formalization corresponds to what is common in Government and BindingTheory, Minimalism, and the less formal variants of Construction Grammar. Asfor the one formal version of Construction Grammar that is a variant of HPSG,namely Sign-Based Construction Grammar (SBCG, Sag 2012), HPSG light can beregarded as a light version of SBCG as well, since the differences are neglectedin the abbreviated representations and trees that are used in this book. The workpresented here differs from non-formal work in GB/Minimalism and Construc-tion Grammar in an important way: it is backed up by implemented grammarsthat use the full version of HPSG including a semantic analysis in the frame-work of Minimal Recursion Semantics (MRS, Copestake, Flickinger, Pollard &Sag (2005)). The detailed analyses are described in conference proceedings, jour-nal articles and books and the reader is invited to consult these resources in caseshe or he is interested in the details. The implemented grammars are distributedwith the Grammix virtual machine (Müller 2007b) and can be downloaded fromthe author’s web-page. The appendix of this book contains a list of sentences thatwere discussed in the respective chapters of this book and which are covered bythe grammars of the respective languages. Readers are invited to enter the sen-tences into the TRALE system (De Kuthy et al. 2004; Penn 2004) that comes with

Preface

Grammix and inspect the complete AVMs.The book starts with two introductiory chapters: the first chapter introduces

the Germanic languages providing basic facts like number of speakers, areaswhere they are spoken, some historical facts. Chapter two discusses the phe-nomena that are treated in the rest of the book, e.g., scrambling, placement ofadverbials, passive, clause types, nonlocal dependencies. The third chapter is anintroduction to Phrase Structure Grammars, which are the foundations of almostall theories since Chomsky’s (1957) formalization of structuralist ideas (Bloom-field 1933). Chapter 3 not just introduces phrase structure grammars but alsogrammars using abstractions over phrase structure rules ultimately resulting invery abstract grammars of the type also known from X theory (Jackendoff 1977).Chapter 4 explains how the concept of valency is combined with abstract phrasestructure rules to make sure that the right number and the right kind of elementsis combined with a certain word. For example, a word like laugh needs a subjectand a word like read needs a subject and an object. This has to be representedsomewhere in a grammar and Chapter 4 explains how it is done in HPSG (light).The basic difference in the analysis of SVO and SOV languages are explained.This chapter also explains how the various orders of subject and objects can beexplained in a language like German (so-called scrambling) and how the variousplacement possibilities in languages like English and the North Germanic lan-guages on the one hand and German, Dutch and Afrikaans on the other handcan be accounted for. Verbal complexes are dealt with in Chapter 5, verb firstposition (used for question formation) and verb second position (for assertions)are explained in Chapter 6. Passive and case assignment in general is treatedin Chapter 7. The Germanic languages are especially interesting here since Ice-landic belongs to this language group and it is known for its quirky subjects(subjects in the genitive, dative, or accusative case, Zaenen et al. 1985). Chapter 8deals with expletive pronouns and how they are used throughout the Germaniclanguages to help marking clause types. For example, expletives are used in Ger-man main clauses to fill the initial position so that the clause is an assertion.Danish uses expletives in embedded sentences with subjects as interrogative el-ements. Again the differences in general grammatical properties influence thegrammar in other parts like placements of expletives.

The final chapter, Chapter 9, is a brief summary of what was done in the bookan points the interested reader to some further literature on HPSG.

German slides developed for the course I am teaching with this book are avail-able on github.1 Lectures in German corresponding to the chapters can also be

1https://github.com/stefan11/germanic-syntax-slides, 2021-09-14.

vi Draft of 20th October 2021, 18:56

https://github.com/stefan11/germanic-syntax-slides

found on youtube.2

On the way this book is published

Teachers at schools and universities are payed by the state, that is by the public(you). Among their duties is the creation of teaching material. There is no reasonwhatsoever to leave the teaching material to profit oriented publishers. On thecontrary, teaching material should be open and adaptable to the needs of theteachers who want to use it.

A study by the American Enterprise Institute shows that the price of collegebooks rose by 812 % from 1978 to 2012 while the general consumer prices rose amere 250 %.3 Similar figures exist for scientific books in general and for universitytext books. My favorite example is a thin text book on logic Logik für Linguisten,which is a translation of the English text book Logic for Linguists (Allwood etal. 1973). This book has 112 pages. It was sold for 9,40e as a paperback by theMax Niemeyer Verlag. This publisher was bought by De Gruyter and the bookis now sold for $126.00/89,95€ as an eBook and $133,00/94,95€ for the hardcoverbook4 (see Müller 2012 for other examples and a general discussion). Both theeBook and the printed book are unaffordable for students. The way out of thishighly problematic situation is to publish books open access. The PDF version ofthis book is free for everybody and the printed copy is available for a reasonableprice since the book is licenced under a Creative Commons license and hence isnot owned by a profit-oriented publisher and everybody can choose his or herown print on demand service in case the default service provided by LanguageScience Press is more expensive.

Gender issues

Macaulay & Brice (1997) did a study on ten textbooks in syntax published be-tween 1969 and 1994. They showed that some of the textbooks used examplesdescribing violence against women. Examples like John beats Mary. or Johnhits Mary. were frequent in the 70ies and 80ies. I know of a paper written bytwo female authors and one male author containing a John hits Mary example.

2https://www.youtube.com/playlist?list=PLXwGGsuPxWRp4AB2LsWH6LKc0II7uc6tg3http://www.aei-ideas.org/2012/12/the-college-textbook-bubble-and-how-the-open-educational-resources-movement-is-going-up-against-the-textbook-cartel/. 10.09.2014.

4http://www.degruyter.com/isbn/978-3-11-096350-2. 2014-09-01.

Draft of 20th October 2021, 18:56 vii

https://www.youtube.com/playlist?list=PLXwGGsuPxWRp4AB2LsWH6LKc0II7uc6tg

http://www.aei-ideas.org/2012/12/the-college-textbook-bubble-and-how-the-open-educational-resources-movement-is-going-up-against-the-textbook-cartel/

http://www.aei-ideas.org/2012/12/the-college-textbook-bubble-and-how-the-open-educational-resources-movement-is-going-up-against-the-textbook-cartel/

http://www.degruyter.com/isbn/978-3-11-096350-2

Preface

Macaulay & Brice (1997: 812) discus some even more extreme cases from thetextbooks they examined. Furthermore, females were depicted in stereotypicalsituations like teaching children, never in work situations with the exception ofwork as secretaries. Some examples were explicitly sexist referring to women asstupid and to men as intelligent.

Macaulay & Brice examined the semantic roles in which women and men ap-peared, checked for usage of full NPs, proper names and pronouns. They foundthat men appear more often than women in example sentences and they appearmore often as agents then as experiencers/patients. While openly sexist exam-ples disappeared from textbooks in the time since the 80ies, Pabst et al. (2017)repeated the studies of Macaulay & Brice and noticed that not much has changedin respect to the women/men distribution in examples.

I think that such biased examples contribute to the stereotypes regarding gen-ders that most of us have since they are deeply entrenched in our societies.Mikaela Wapman and Deborah Belle did a study asking students of psychologyabout the following situation: a father and son are in a horrible car crash andthe father gets killed. The son is brought to the hospital and has to undergo anemergency surgery but the surgeon refuses to do it and say, “I can’t operate—that boy is my son!”. Question: How is this possible? People came up withall sorts of explanations like gay fathers, ghost surgeons and so on, overlook-ing the possibility of the surgeon being the mother of the child. This showsthat the idea that surgeons are male is deeply entrenched in our societies (thesame experiment works in German and it works in the other direction withnurses). The obvious solution to such problems is to change the employmentof women and to change the payment of jobs usually taken by women. This isa political problem and it would be helpful if women were in at least 50 % ofthe positions in political systems. Many of the problems that women have canbe traced back to a lack of economic independence. There is a good portrayof East German women showing the consequences of economic independence:https://www.mdr.de/zeitreise/schwerpunkte/ostfrauen-106.html. When the wallcame down and East German women andWest German women united there wasmuch confusion since the issues they wanted to address were totally different(see also the MDR video documentation). Almost all women in East Germanyworked, child care was available everywhere and so on.

What we seem to have here is an hen egg problem: society and stereotypeswould be different if women would be in other jobs and better payed, but stereo-types prevent women from even trying because they feel not welcome in certainjobs. Here is where language plays an important role. German marks gender for

viii Draft of 20th October 2021, 18:56

https://www.mdr.de/zeitreise/schwerpunkte/ostfrauen-106.html

professions: one can talk about Kindergärtner and Kindergärtnerinnen. Lookingat the German grammatical system, the plural is unmarked for gender, but thereis the form Kindergärtnerinnen that makes gender explicit. Studies show thatrecipients do not consider women if the unmarked form is used. For instance,Stahlberg et al. (2001) showed that people come up with more males when askedfor their favorite Musiker rather than their favorite Musiker ‘musician’ or Musik-erin ‘female musician’. Germans solved this problem by using abbreviations forfusing both variants: MusikerIn. Further more inclusive forms have been devel-oped including people with diverse genders: Musiker*in. These forms are stan-dard in the communication of the German Research Foundation and some univer-sities (e.g., the Humboldt-Universität zu Berlin, where I am currently working).

While the so-called Binnen-I helps to make women visible in communicationthis does not help as far as textbooks are concerned since respective forms arerather rarely used in examples. So, to make non-male readers feel more wel-come, other means have to be employed. One has to work on the referents thatare used in examples. The names used most frequently in examples in the pa-pers/textbooks from which I was learning syntax (written in the 80ies and 90ies)were John andMary or Karl andMaria in German publications. (Books on HPSGusing gender neutral names like Kim and Sandy were an early exception (Pollard& Sag 1987; 1994).) I never used discriminating examples in my publications and Iavoided stereotypes, but I also used Karl andMaria and I had den Mann ‘the man’reading books and der Mann ‘the man’ giving books to der Frau ‘the woman’ ordem Kind (neuter, ‘the child’). One reason for having more male referents in Ger-man examples is that the masculine inflection paradigm differentiates betweenall four cases while the feminine one collapses nominative and accusative as wellas genitive and dative. The neuter is somewhere in the middle: nominative andaccusative are syncretic. Table 1 gives an overview of the situation. If one wants

Table 1: Inflectional marking within noun phrases of feminine and masculinegender. Only themasculine paradigm is unambiguouslymarked for case.

the man the woman the child

nominative der Mann die Frau das Kindgenitive des Mannes der Frau des Kindesdative dem Mann der Frau dem Kindaccusative den Mann die Frau das Kind

to avoid male referents, one could use animals, but depending on the verb one

Draft of 20th October 2021, 18:56 ix

Preface

uses, a dolphin as subject would not make sense. So, one way to solve this prob-lem, is to use proper names like Kim, Sandy, and Chris that are unisex. This isof course only possible if case is unimportant since proper names do not inflectfor case in almost all of the Germanic languages (Icelandic being the exception,Zaenen et al. 1985: 443).

As mentioned above there is some tradition of what names are used in lin-guistic examples: for English it was John and Mary, for German Karl and Maria,Dutch authors used Jan and Piet and Marie. Since this book is about Germaniclanguages I thought I collect names that are typical for the languages under dis-cussion. It is clear that the names trick does not work globally. I learned that Gertcan be used both for women and men in the Scandinavian countries, but in Ger-many this name exclusively refers to males.5 In addition speaker communitiesmay associate certain names with women and girls since there are more femalesthan males with this name in a certain community and vice versa. And finallythere are differences on the individual level: you may happen to know three fe-male Connys and no male one but this may change during the course of yourlive. So an example about Conny reading a book or feeding a child could workfor or against a certain stereotype, depending on the experiences of the reader.And of course the assignment to a certain gender may change in time: a namethat is given more often to girls may be given more often to boys some decadeslater. So, what I did is not perfect but since I state here that all names are meantto be unisex including non-binary people, it is a step into the right direction.

In a discussion on twitter, XY pointed out to me that names do not necessarilyfit the gender of the referent. There are famous examples of a male US Americanlawyer with the name Sue6 and of women with the name Michael (the columnistMichael Sneed, Chicago Sun-Times, and The Bangles’ bassist Michael Steele).7

So in principle one would have to make the gender of the referent explicit. Butthis is exactly what I want to avoid by using unisex names: I do not want toentrench stereotypes. By not using Sue, I do not have to say that Sue may be aman.

A special case are non-binary people. They may not feel represented in ex-amples with unisex names like Kim and Sandy. This is the reason why theysometimes choose names that are not used by binary people. Kirby Conrod sug-

5See Bernadette La Hengst’s song Ein Mädchen namens Gerd ‘A girl named Gerd’, which is a par-allel toA boy named Sue by JohnnyCash: https://www.youtube.com/watch?v=Y6WVcOk4D3c,2020-06-26.

6https://en.wikipedia.org/wiki/Sue_K._Hicks, 2020-07-06.7https://en.wikipedia.org/wiki/Michael_Burnham, 2020-07-06.

x Draft of 20th October 2021, 18:56

https://www.youtube.com/watch?v=Y6WVcOk4D3c

https://en.wikipedia.org/wiki/Sue_K._Hicks

https://en.wikipedia.org/wiki/Michael_Burnham

gested that non-binary people give permission to use their names and allowedme to use theirs (Kirby’s), which I do at some places in the book.

Table 2 shows the names I am using throughout the book.

Table 2: This table lists the names used in the book for the languages given

language name female basis male basis

Danish Gert/Gerd Asgjerd, Hallgerd, Hilde-gard, Ingjerd

Gerhard

Dutch Robin Robrecht, RobertBenedikt

German Conny Constanze, Cornelia Conrad, ConstantinAicke Eckehard

English Chris Christina, Christine Christian, Christoph(er)Kim Kimberly Kimberly, Kimball,

Joachim, JoakimSandy Sandra Alexander, Sander, Alas-

dairnon-binary Kirby

Elvan

Berlin, 20th October 2021 Stefan Müller

Draft of 20th October 2021, 18:56 xi

Acknowledgments

I thank Hubert Haider and Matthias Hühning for discussion.I want to thank all students of my course on the structure of Germanic lan-

guages. This book benefited enourmously from teir questions and the discussionduring the lectures. Carolin Ulmer, and Bastian Schoenfeld deserve special men-tion for important comments. I thank Sarah Böke, Alexandra Fosså, Anne Kilgus,Nils P. Kujath and Maya Tabea Wolff for pointing out typos.

I thank the twitter community for discussing gender issues and names in lin-guistics examples withme. Special thanks go to LaurenAckerman, Kirby Conrod,and Brianna Wilson for extensive discussion.

Abbreviations

PART particlePREP preposition

1 A general overview of the Germaniclanguages

This chapter provides an overview of general facts about theGermanic languages.It derives from slides for teaching courses about Germanic languages that wereused by Ekkehard König and passed on to Matthias Hüning and via Matthias tome, which explains the similarity to the introductionary chapter by Henriksen& van der Auwera (1994) in the book The Germanic Languages edited by König& van der Auwera (1994).

1.1 Languages and speakers

Depending on whom one asks there are app. 5000 to 6000 languages spokenworldwide currently. The Germanic languages are a small subset (depending onthe counting 15 languages). The problem is the distinction between language andlanguage variety (e.g., varieties of Frisian). According to Max Weinreich a lan-guage is a dialect with an army and a navy. According to this “definition”, SwissGerman would not be a language. That the Swiss army has a bicycle group in-stead of a navy does not help. This brief discussion should indicate that it is oftena political question whether two variants of a language are treated as differentlanguages or not (Slovak vs. Czech and Serbian vs. Croatian). Altogether thereare almost 500 million native speakers, which is 1/12 of the whole population ofthe world. Especially English is wide spread in terms of regions in which thelanguage is spoken.

1.2 Historical remarks and relatedness between thelanguages

The Germanic languages constitute a separate branch of the tree representingthe indo-european language family (see Figure 1.1 on the following page). Proto-Germanic formed between 2000 and 1000 BC. Its origins are in the Baltic region,

1 A general overview of the Germanic languages

Figure 1.1: Language tree according to Fitch (2007: 665)

2 Draft of 20th October 2021, 18:56

1.2 Historical remarks and relatedness between the languages

that is, in northern Germany and south Scandinavia. The area where it wasspoken extended from the North Sea to Poland (app. 500 BC).

First Germanic Sound Shift took place app. until the second century BC. Inthat millenium the Germanic languages developed different consonants from theother indo-european languages. The first written documents are runes from app.300 AC and the Gothic Bibel translation in the fourth century.

Wikipedia1 provides the table in Figure 1.2 on the next page that depicts thedevelopment of the Germanic languages. Germanic is devided into East, West,and North Germanic. East Germanic existed in the form of Gothic until app.1800 on the Crim, as Crimean Gothic and is now totally extinct. West Germanicconsists of

• German,

• Yiddish,

• Luxembourgish,

• Pennsylvania Dutch,

• Low German,

• Plautdietsch (also called Mennonite Low German),

• Dutch,

• Afrikaans,

• Frisian, and

• English.

The North Germanic languages are:

• Danish,

• Swedish

• Norwegian

• Icelandic, and

• Faroese.

Table 1.1 on page 5 shows how similar the words from the main vocabulary ofthe Germanic languages are:

1https://en.wikipedia.org/wiki/Germanic_languages. 19.10.2014

Draft of 20th October 2021, 18:56 3

https://en.wikipedia.org/wiki/Germanic_languages


Figure 1.2: History and grouping of Germanic languages according to Wikipedia


1.3 The three Germanic branches

Table 1.1: Words from the main vocabulary of some Germanic languages

Dutch vader vier vol huis bruin uit kruid muisGerman Vater vier voll Haus braun aus Kraut MausEnglish father four full house brown out crowd (?) mouseFrisian – fjouwer fol hûs brún út krûd mûsSwedish fader fyra full hus brun ut krut musDanish fader fire fuld hus brun ud krudt musNorwegian far fire full hus brun ut krydder musIclandic faðir fjórir fullur hús brúnn út – mús


Proto-Germanic developed into the three main branches East, West, and NorthGermanic, approximately in the first centuary AD. The reasons for this devel-opment were inherent variations in the respective dialects, migration (languagecontact) and standardization. This book treats the structure of the Germanic stan-dard languages. This section is devided into three subsections that correspond tothe three main Germanic branches. I will sketch over the historic developmentsthat lead to the languages spoken today. Many of the details that are covered inFigure 1.2 will be ignored.

1.3.1 East Germanic

The Goths emigrated from the Danish islands and South Sweden approximately100 BC and meet the Vandals and other tribes. Gothic and Burgundian and somesmaller languages constitute the East Germanic branch, of which only Gothic gotpassed on. After the decay of the Gothic empires Gothic died out. There weresome remains on the peninsula Crimea until app. 1800. The west gothic bishopWulfila translated the Bible into Gothic. The best-known version of it is thefragment Codex Argenteus, which belongs to the university library of Uppsala.Figure 1.3 on the next page shows a picture of it.2

1.3.2 North Germanic

The first writings on runestones date back to the 6th century. The language of theWikings (800–1050) was rather homogenous and it was only after this era that

2Taken from Wikipedia: http://de.wikipedia.org/wiki/Bild:Wulfila_bibel.jpg. 19.10.2014.


http://de.wikipedia.org/wiki/Bild:Wulfila_bibel.jpg


Figure 1.3: The Wulfila Bible (Codex Argenteus), picture from Wikipedia

two branches were starting to develop: the east-scandinavian branch with Old-Danish and Old-Swedish and the West-Scandinavian one with Old-Norwegianand Old-Icelandic.

1.3.2.1 Danish

Danish (dansk) is the official language of the Kingdom of Denmark, second of-ficial language of the Faroe Islands and of Greenland, Inuit being the first offi-cial language of Greenland. Danish has app. 5,5 million speakers. About 50,000speakers live in Schleswig-Holstein, the northernmost of the federal states ofGermany. Danish is the Scandinavian language that drifted furthest away fromthe common scandinavic roots.

1.3.2.2 Swedish

Swedisch (svenska) is the official language in Sweden with app. 8,5 million na-tive speakers. It is the first language of app. 300,000 Swedish-speaking Finns inFinnland. Until the times of the Wikings Danish and Swedish were almost in-distinguishable. Starting from app. 800 they started to diverge. Since app. 1300there are obvious differences.


https://en.wikipedia.org/wiki/Denmark

https://en.wikipedia.org/wiki/Schleswig-Holstein


1.3.2.3 Iclandic

Icelandic (íslenska) is the west-scandinavian language of Iceland since the set-tlement over 1000 years ago. There are app. 325,000 native speakers. 97 % of theIcelandic population (325,000) has Icelandic as their mother tongue and there arelarger groups of native speakers in Denmark, the USA, and Canada (app. 15,000in total). There is almost no variation (no dialects). The language is conserva-tive, in the sense that Icelandic is the language among the Germanic languagesthat best preserved the Germanic vocabulary and inflection. In the beginningthere were almost no differences between Norwegian and Icelandic but then thelangauges diverged.

1.3.2.4 Norwegian

There are two varieties of Norwegian (norsk): Danish-Norwegian (bokmål) andNew-Norwegian (nynorsk). Both are official languages of Norway and are usedin parallel. There are app. 4,3 million speakers. From 1380 to 1814 Danish was thewritten language and local dialects were spoken. Hence, a Norwegian standardhad to be developed. This was done by Ivar Aasen (1813–1896), who developedNynorsk. Nynorsk got an official status in 1885. Bokmål ‘book tongue’ is the firstlanguage of most of the Norwegians.

1.3.2.5 Faroese

Faroese (føroyskt) is – together with Danish – the official language of the FaroeIslands. There are app. 47,000 speakers. The Faroe Islands belong to Denmarksince 1816. Since 1948 they are a self-governing country within the Danish Realm.Faroese has a strong Danish influence. The first manuscript transmission is asrecent as 1773 and even then there are not many written documents (in contrastto Icelandic).

1.3.3 West Germanic

Opinions on the question whether West-Germanic did develop from a singlesource or not differ. Some authors assume that the West Germanic languagesdo not have a common root, but instead developed from the following three un-related branches of dialect groups (for instance Robinson 1992: 17–18 and Hen-riksen & van der Auwera 1994: 9):

• North Sea Germanic


https://en.wikipedia.org/wiki/Danish_Realm


• Weser-Rhine Germanic

• Elbe Germanic

Other authors assume that these three branches had a common ancestor (seeFigure 1.2).

There is no unique mapping of these dialect groups to the languages spokentoday.

1.3.3.1 German

German is the official language of

• Germany (app. 80 million speakers),

• Austria (app. 7,5 million speakers),

• Liechtenstein (app. 15,000 speakers),

• Switzerland (4,2 million of 6,4 million Swiss residents),

• Northern Italy/South Tyrol (app. 270,000 speakers),

• Belgium (app. 65,000 speakers), and

• Luxembourg (app. 360,000 speakers).

Luxembourg has next to German also the original language Luxembourgish andFrench as official languages.

There are app. 97 million speakers of German, app. 90 of those are nativespeakers and 7 million have German as their second language. Approximately80 million speakers speak German as a foreign language, app. 55 Milion of theselive in the European Union.

There are three national main variants (Germany, Austria, Switzerland). Inother countries German is usually a minority language. There are two largerdialect groups: Low German (Plattdüütsch, Nedderdüütsch; Standard German:Plattdeutsch or Niederdeutsch; Dutch: Nedersaksisch in the wider sense) andHigh German (varieties of German spoken south of the Benrath and Uerdingenisoglosses).



1.3.3.2 Yiddish

Yiddish is one of many Jewish languages. App. 2 million people speak this lan-guages in various regions of the world, most of them in the USA (1,25 Mill.). Thenumber of Yiddish speakers was much higher 100 years ago: 7 million speakersof Yiddish lived in Europe, most of them in Russia and in Austria-Hungary. Atmost 75,000 Yiddish speakers are left in Western Europe. Yiddish has its roots inmedieval German with influences from Hebrew and Aramaic.

1.3.3.3 Pennsylvania German

Pennsylvania German (Pensilfaanish, Deitsch), which is also known as Pennsyl-vanian Dutch has app. 300,000 native speakers, who mainly live in the USA. Themost important regions are Pennsylvania, Ohio, and Indiana. Pennsylvania Ger-man is the result of immigration in the 17th and 18th century. Members of vari-ous protestant religions (Mennonites, Pietists and so on) left Europe for reasonshaving to do with their religions but later immigrants that came for economicreasons followed. The language is based on Palatine dialects and is nowadaysmainly spoken by Amish and Mennonites.

1.3.3.4 Dutch

Dutch (Nederlands) is the official language of the Netherlands and has app. 15million speakers there. Dutch is one of the official languages of Belgium (app.6 million speakers; almost 4 million Walloones). Dutch is the sole official lan-guage and teaching language in Suriname, which is independent since 1975, andin Aruba and the Netherlands Antilles.

1.3.3.5 Afrikaans

Afrikaans is one of the official languages of South Africa since 1925, which haseleven official languages. There are app. 6,4 million native speakers in SouthAfrica, which is about 15 % of the population and 150,000 in Namibia. Afrikaansdeveloped since the 17th century from Dutch dialects and is seen as an indepen-dent language since somewhere between 1775 and 1850 (den Besten 2012). Thevarious languages spoken in the area interacted with Afrikaans and its originsand resulted in a structural simplification in comparison to Dutch. Today Englishhas a strong influence.


https://en.wikipedia.org/wiki/Aruba

https://en.wikipedia.org/wiki/Netherlands_Antilles


1.3.3.6 Frisian

The three varieties of Frisian are mutually not intelligible. There is North Frisianspoken by app. 10,000 speakers mainly on the north Frisian islands Amrum, Sylt,and Helgoland. East Frisian is extinct with the exception of Saterlandic, which isspoken in the three villages of Saterland (Landkreis Cloppenburg ) by between1,000 and 2,500 people. West Frisian is spoken in the northern Dutch provinceFryslân (Friesland) and has app. 350,000 native speakers.

1.3.3.7 English

English has app. 570 million speakers at the end of the 20th century all overthe world (337 million native speakers, 235 million speakers with English as thesecond language). The following list provides the countries with the most nativespeakers:

• USA: 227 million,

• Great Britain: 57 million,

• Nigeria: 43 million,

• Canada: 24 million,

• Australia: 17 million,

• Ireland: 3,5 million,

• New Zealand: 3,2 million.

There aremany national variants, which differmostly in pronunciation. Between1 and 1,5 billion people have active or passive knowledge of English. English isofficial language in 59 states. It is the most important scientific language.


2 Phenomena

This chapter deals with variation in the Germanic languages in what is oftencalled the Core Grammar, that is in sentences of the John loves Mary variety.1

We will look at differences in the verb position (verb before object and objectbefore verb), the verb second property, which all of the Germanic languages withthe exception of English have, the ordering of subjects and obejcts with respectto each other, the placement of adverbials, the existence/non-existence of verbalcomplexes, the obligatoriness/absence of subjects, passive including the personaland impersonal passive, expletive pronouns and various ways to mark the clausetype, that is, to signal whether a certain clause is an assertion, a question or anembedded clause.

A note of caution is necessary here: especially the following three subsectionsare potentially confusing. A language like German will be categorized as anSubject-Object-Verb language, a verb-second language and a language with freeconstituent order (Haftka 1996). This sounds contradictory but it is not. Therespective classifications refer to properties of languages as such not to the formof single sentences.

2.1 Order of subject, object and verb

The langauges of the world can be classified according to the order of subject,object, and verb that is dominant (Greenberg 1963). In order to make languagescomparable a very general definition of grammatical functions like subject andobject is used for such a classification. The definition is based on semantic prop-erties: subjects are those arguments that are agentlike and objects are argumentsthat are rather patient like. This definition is not always identical to the language-particular definitions. For instance, if one follows the semantic definition, the

1Chomsky (1981: 7–8) suggests dividing grammars of natural languages into a core part and aperiphery. All regular parts belong to the core. The core grammar of a language is assumedto be an instance of Universal Grammar (UG), the genetically determined innate languagefaculty of human beings. Idioms and other irregular parts of a language belong to the periph-ery. This book deals with phenomena usually assumed to be the core without assuming thiscore/periphery distinction and without assuming an UG (Müller 2014; 2015c).

2 Phenomena

phrase der Aufsatz ‘the paper’ in (1) is the subject although it is inanimate andnot an agent:

(1) Derthe

Aufsatzpaper

interessiertinterests

mich.me

‘I am interested in the paper.’

The language-particular definition of subject in German (and the Germanic lan-guages in general) rather refers to properties like nominative case, subject-verb-agreement, and control. We will deal with this in more detail in Section 7.1.1.

Figure 2.1 shows the dominant order of subject, object, and verb among theworld’s languages. According to Dryer (2013a) the dominant order is defined asfollows:

Where a language is shown on one of the word order maps as having aparticular order as the dominant order in the language, this means that itis either the only order possible or the order that is more frequently used.(Dryer 2013a)

Figure 2.1: Dryer (2013b: Section 1): Feature 81A: Order of subject, object andverb, The World Atlas of Language Structures

If we zoom in to display the European languages we get Figure 2.2 on thenext page. According to the WALS the languages Iclandic, Norwegian, Swedish,Danish, and English are SVO languages. Dutch, German, and Frisian, however,are marked in grey, that is, these languages are marked to have no dominant



Figure 2.2: Dominant orders of subject, object, and verb in Europe

order.2 According to Figure 2.3 on the following page these languages have twodominant orders, namely SOV and SVO. The reason for this classification is thatDryer (2013b: Section 1) distinguishes between sentences in which the finite verbis the main verb (2a) and sentences in which the finite verb is an auxiliary as in(2b):

(2) a. KimKim

siehtsees

denthe

Fuchs.fox

‘Kim sees the fox.’b. Kim

Kimhathas

denthe

Fuchsfox

gesehen.seen

‘Kim has seen the fox.’

According to Dryer the pattern for (2a) is SVO and the one for (2b) is SAuxOV,where Aux stands for the auxiliary verb. Like Greenberg (1963)3, Dryer countsthe latter pattern as SOV order. The question is whether it is adequate to ignore

2Greenberg (1963: 87) listed German and Dutch among the SVO languages.3See Höhle (2019c: 20–21).


2 Phenomena

Figure 2.3: Two dominant orders of subject, object, and verb (Dryer 2013b: Sec-tion 3)

auxiliaries in the examination of constituent order. The auxiliary hat ‘have’ in(2b) syntactically behaves like the full verb scheint ‘seems’ in (3):

(3) KimKim

scheintseems

denthe

Fuchsfox

zuto

sehen.see

‘Kim seems to see the fox.’

So, here we would have an SVOV order, something that does not exist in thetypology under discussion. The languages in Figure 2.3 marked as not having adominant order use the verb position to mark the clause type: it is just the finiteverb that is in first or second position. Non-finite verbs are final:

(4) KimKim

scheintseems

denthe

Fuchsfox

gesehenseen

zuto

haben.have

‘Kim seems to have seen the fox.’

In subordinate clauses we have both the finite verb and the non-finite verbs infinal position while we have the finite verb in initial position4 in questions and

4I use the term initial position to refer to the position the finite verb has in V1 or V2 clauses.



declarative main clauses. A classification that is entirely based on counting pat-terns without taking auxiliary verbs and the finiteness/non-finiteness distinctioninto account cannot tear these properties apart (Höhle 1983: Section 3). In whatfollows we will have a look at clauses with both finite and non-finite verbs. Suchclauses reveal differences between OV and VO languages and I will argue thatAfrikaans, Dutch, German, and Frisian should be counted among the OV lan-guages and that the other observable pattern SVO is due to other properties ofthese languages, namely that they mark the clause type by verb position and thatthey are verb second (V2) languages.5

When one builds more complex German sentences involving several verbs,the embedding verb is usually realized to the right of the embedded verbs. Thisis shown in (5). (5a) shows a simple sentence with a finite verb. If we form theperfect as in (5b), the perfect auxiliary has to follow the participle. The auxiliaryis the finite verb and it determines the form of the participle. Hence the finiteverb is the verb that embeds the participle. This is indicated by the lower numberof hat in comparison to gesehen. If we build an even more complex sentence byadding another verb, this verb will be serialized to the right of the present verbs(5c). The Danish example in (6c), quoted from Ørsnes (2009: 146), corresponds tothe German example in (5c).

(5) a. dassthat

sieshe

ihnhim

sieht1sees

(German)

‘that she sees him’b. dass

thatsieshe

ihnhim

gesehen2seen

hat1has

‘that she has seen him’c. dass

thatsieshe

ihnhim

gesehen3seen

haben2have

muss1must

‘that she must have seen him’

(6) a. atthat

hunshe

ser1sees

hamhim

(Danish)

b. atthat

hunshe

have1has

set2seen

hamhim

The analysis of V2 and V1 involves fronting of the finite verb. In V2 clauses a constituent isfronted in addition.

5The property of being a V2 language is independent of the SVO/SOV distinction. All Germaniclanguages except English are V2 languages. See Section 2.2 on V2.


2 Phenomena

c. atthat

hunshe

må1must

have2have

set3seen

hamhim

As the examples in (6) show, the verbs are added in front of the verbs they embedin Danish. This is also the case for English as is evident from the glosses. InDanish and English the verbs precede the object (ham/him) and in German theyfollow it (ihn).

Haider (2020: Section 15.2) pointed out two further differences between theGermanic VO and OV languages: particles precede verbs in OV languages andthe same is true for resultative secondary predicates. In VO languages particlesand result predicates follow the verb. This is demonstrated by the following twoexample sets:

(7) a. Kim will look up the information.b. Kim

Kimwirdwill

diethe

Informationinformation

nachschlagen.PART.beat

(German)

‘Kim will look up the information.’

(8) a. Kim will fish the pond empty.b. Kim

Kimwirdwill

denthe

Teichpond

leerempty

fischen.fish

(German)

(7a) shows that look precedes the particle up, while the verb schlagen ‘beat’ has tofollow the particle nach in German. Similarly, the secondary resultative predicateempty follows the verb in (8a), but leer precedes the verb in (8b). Note that I useda future auxiliary in the examples in order to avoid side effects that are due tothe verb second property of German: in declarative main clauses the finite verbalways precedes particles and resultative predicates but this is due to the clausetype (see Section 2.2).

I will return to the SVO vs. SOV order in Chapter 6 and provide more evi-dence that was used in the literature to argue for the OV status of languages likeGerman and Dutch.

2.2 V2

The Germanic languages, with the exception of English, are so-called verb secondlanguages (V2 languages). The V2 property can be illustrated with the followingGerman sentences. (9) shows declarative main clauses in which one of the con-stituents is fronted. (10) shows parallel interrogative clauses.


2.2 V2

(9) a. Dasthe

Kindchild

gibtgives

demthe

Eichhörnchensquirrel

jetztnow

einea

Nuss.nut

(German)

‘The child gives the squirrel a nut now.’b. Dem

theEichhörnchensquirrel

gibtgives

dasthe

Kindchild

jetztnow

einea

Nuss.nut

c. Einea

Nussnut

gibtgives

dasthe

Kindchild

demthe


jetzt.now

d. Jetztnow

gibtgives

dasthe

Kindchild

demthe


einea

Nuss.nut

(10) a. Werwho

gibtgives

demthe


jetztnow

einea

Nuss?nut

(German)

‘Who gives the squirrel a nut now?’b. Wem

whogibtgives

dasthe

Kindchild

jetztnow

einea

Nuss?nut

‘Who does the child give a nut to now?’c. Was

whatgibtgives

dasthe

Kindchild

demthe


jetzt?now

‘What does the child give the squirrel now?’d. Wann

whengibtgives

dasthe

Kindchild

demthe


einea

Nuss?nut

‘When does the child give the squirrel a nut?’

The finite verb is in second position in all the sentences in (9) and (10).English, in contrast, does not allow orders in which the object appears imme-

diately before the finite verb.

(11) a. * This squirrel give I a nut now.b. * This nut give I a squirrel now.c. * Tomorrow give I the squirrel a nut.

Adverbials and objects can be fronted but then they have to appear before theclause consisting of subject and verb and possibly other constituents.

(12) a. This nut, I give the squirrel now.b. Now, I give the squirrel a nut.


2 Phenomena

Note also that fronting of objects is restricted to the secondary object for verbswith two objects for some speakers (Hudson 1992: 258).6 So, while fronting of thesecondary object in (13b) is permitted by all speakers, some speakers find extrac-tions like the extraction of the primary object in (13c) unacceptable or marked.

(13) a. We give children sweets.b. These sweets, we give children _.c. % These children, we give _ sweets.

This is not the case in V2 languages: they are rather liberal as far as fronting isconcerned. Basically all constituents can be fronted, exceptions being reflexivepronouns that are selected by inherently reflexive verbs (14), expletive objects(15), and certain modal particles (16). See also Hoberg (1981: 159) on inherentlyreflexive verbs and modal particles.

(14) a. MariaMaria

erholtrecovers

sich.REFL

(German)

‘Maria recovers.’b. * Sich

REFLerholtrecovers

Maria.Maria

(15) a. Erhe

bringtbrings

esEXPL

bisuntil

zumto.the

Professor.professor

(German)

‘He makes it to professor.’b. # Es

EXPLbringtbrings

erhe

bisuntil

zumto.the

Professor.professor

(16) a. Erhe

gehtgoes

haltPARTICLE

nicht.not

(German)

‘He simply does not go.’b. * Halt

PARTICLEgehtgoes

erhe

nicht.not

The element in front of the finite verb is not necessarily a clause mate of thefinite verb. In fact, it can belong to a deeply embedded head as is demonstratedby the following example from German:

6I use the terms primary object and secondary object in order to avoid confusion that is some-times caused by the terms direct and indirect object. The primary object is the first objectin English and the dative object of ditransitive verbs governing the dative in German. Thesecondary object is the second object in English and the accusative in German ditransitiveconstructions.


2.2 V2

(17) [Überabout

diesesthis

Thema]𝑖topic

habehave

ichI

sieher

gebeten,asked

[[einena

Vortragtalk

_𝑖 ] zuto

halten]?7

hold(German)

‘I asked her to give a talk about this topic.’

The PP über dieses Thema depends on Vortrag ‘talk’, which is part of the VPheaded by zu halten ‘to hold’, which is in turn embedded under gebeten ‘asked’.Sentences like (17) show that V2 frontings cannot be analyzed as a simple re-ordering of the arguments of a verb. While such an approach would work forthe examples in (18), it would not extend to other cases in which the frontedelement does not depend on the highest verb in the clause.

(18) Denthe.ACC

Texttext

kenntknows

er.he

‘He knows the text.’

The following examples from Danish (SVO) show that the property of being aV2 language is independent of the VO/OV property:

(19) a. GertGert

harhas

læstread

bogen.book.DEF

(Danish)

b. Bogenbook.DEF

harhas

GertGert

læst.read

The example in (19a) shows that the object follows the verbs and (19b) shows thatthe object bogen ‘the book’ can appear in sentence initial position infront of thefinite verb har ‘have’.

The V2 order is used in declarative main clauses throughout the Germaniclanguages (without English). Some Germanic languages do not use V2 order inembedded clauses. For a discussion of embedded interrogatives see Section 2.5.

While English does not allow for the order object verb subject, which is possi-ble in the other Germanic languages due to V2 fronting, it allows for the frontingof the object in questions resulting in structures that are parallel towhatwe knowfrom the other Germanic languages:

(20) a. Which book did Sandy read?b. Which book did Sandy give to Kim?c. To whom did Sandy give the book?

7Adapted from Hinrichs & Nakazawa (1989a: 21).


2 Phenomena

English used to be a V2 language but lost this property. The V2 in questions is aresidue of earlier stages of the language, which is why English is called a residualV2 language (Rizzi 1990: 375).

V2 and verb fronting in general is a way to mark clause types in all Germaniclanguages. V2 sentence can be declarative clauses in all Germanic languagesexcept English and they can be questions in all Germanic languages includingEnglish. In addition V2 sentences may be imperatives, as (21) shows.

(21) Jetztnow

gibgive

ihrher

dasthe

Buch!book

(German)

‘Give her the book now!’

Sentences with the finite verb in first position (V1) can be yes/no questions orimperatives:

(22) a. Gibtgives

erhe

ihrher

dasthe

Buch?book

(German)

‘Does he give her the book?’b. Gib

giveihrher

dasthe

Buch!book

Of course the order of elements is not the only cue as far as the clause type isconcerned. Intonation and morphological marking of imperative forms plays arole as well.

The property of being a V2 language is exceedingly rare among the world’slanguages (Holmberg 2015: 343). Apart from the Germanic languages with theexception of Modern English (Haider & Prinzhorn 1986), there are only Esto-nian (Finno-Ugric, Holmberg 2015: 343), Sorbian (Plank 2003: entry 79) (a Slaviclanguage), the Celtic languages Breton (Borsley & Kathol 2000), Cornish (Bors-ley, Tallerman & Willis 2007: 287), and Middle Welsh (Willis 1998), Old French(Adams 1987: Section 1.3; Roberts 1993: Section 2.1.2; Vance 1997: Chapter 2), OldSpanish (Fontana 1997: Section 3.3.2), Rhaeto-Romance (Poletto 2002; Anderson2006), Kashmiri (Bhatt 1999: Chapter 4), two dialects of Himachali, which alsobelongs to Indo-Aryan and is spoken in regions adjacent to Kashmiri (Hendrik-sen 1990), the Austronesian languages Taiof and Sisiqa (Ross 2004: 495) and theBrazilian native language Karitiana from the language family Tupí (Storto 2003).


2.3 Scrambling

2.3 Scrambling

While the constituent order in languages like English is rather fixed, languageslike Dutch and German allow a rather free permutation of arguments. In ordernot to contaminate the effects by reorderings that are due to the V2 property, I useverb last sentences to illustrate the phenomenon. Example (23) shows the onlypossible order for subject and objects of a simple ditransitive sentence withoutextraction:

(23) because the child gives the squirrel the nut (English)

If speakers want to realize the secondary object the nut before the primary objectthe squirrel, they have to use a prepositional object. This type of reordering iscalled dative-shift and an example is provided in (24):

(24) because the child gives the nut to the squirrel (English)

In contrast to this we have the German examples in (25). These examples showthat the noun phrases can be freely permuted:

(25) a. [weil]because

dasthe

Kindchild

demthe


diethe

Nussnut

gibtgives

(German)

b. [weil]because

dasthe

Kindchild

diethe

Nussnut

demthe


gibtgives

c. [weil]because

diethe

Nussnut

dasthe

Kindchild

demthe


gibtgives

d. [weil]because

diethe

Nussnut

demthe


dasthe

Kindchild

gibtgives

e. [weil]because

demthe


dasthe

Kindchild

diethe

Nussnut

gibtgives

f. [weil]because

demthe


diethe

Nussnut

dasthe

Kindchild

gibtgives

Not all of these orders can be used in all contexts. Some of the examples requirea special, contrastive intonation. The orders can be sorted with respect to thenumber of contexts in which they can be used. Höhle (1982) suggests calling theorder that can be used in most contexts the normal or unmarked order.


2 Phenomena

2.4 The position of adverbials

In languages like German and Dutch, the position of adverbials is rather free: theadverb gestern ‘yesterday’ can appear anywhere between the arguments and theverb:

(26) a. weilbecause

dasthe

Kindchild

demthe


diethe

Nussnut

gesternyesterday

gabgave

(German)

‘because the child gave the squirrel the nut yesterday’b. weil

becausedasthe

Kindchild

demthe


gesternyesterday

diethe

Nussnut

gabgave

c. weilbecause

dasthe

Kindchild

gesternyesterday

demthe


diethe

Nussnut

gabgave

d. weilbecause

gesternyesterday

dasthe

Kindchild

demthe


diethe

Nussnut

gabgave

Dutch has free order of adverbials as well (Koster 1999: 4; Bouma 2003: Section 6).(27) shows the Dutch examples corresponding to (26):

(27) a. omdatbecause

hetthe

kindchild

dethe

eekhoornsquirrel

dethe

nootnut

gisterenyesterday

gafgave

(Dutch)

‘because the child gave the squirrel the nut yesterday’b. omdat

becausehetthe

kindchild

dethe

eekhoornsquirrel

gisterenyesterday

dethe

nootnut

gafgave

c. omdatbecause

hetthe

kindchild

gisterenyesterday

dethe

eekhoornsquirrel

dethe

nootnut

gafgave

d. omdatbecause

gisterenyesterday

hetthe

kindchild

dethe

eekhoornsquirrel

dethe

nootnut

gafgave

In contrast, the position of the adverbials is rather restricted in SVO languageslike Danish and English. The adverbials usually are placed before or after the VP;that is, verb and objects form one unit and adverbials attach to the left or to theright of this unit. (28) provides an example:

(28) a. because the child often [gave the squirrel the nut] (English)b. because the child [gave the squirrel the nut] oftenc. * because the child [gave often the squirrel the nut]d. * because the child [gave the squirrel often the nut]


2.5 Embedded clauses

It is assumed that verb and objects form a structural unit, a verb phrase (VP).Adverbials may attach to this VP forming a larger VP, which is than combinedwith the subject to form a complete sentence.

The following example, which is due to Quirk et al. (1985: § 8.20, 495), showsthat even in very complex combinations of several verbs adverbs may be placedat the left periphery of a VP:

(29) It [certainly [VP may [possibly [VP have [indeed [VP been[badly [VP formulated]]]]]]]].

This is different from theOV languageswhere verbs form a verbal complexwhichusually cannot be interrupted by adverbs.

(30) a. dassthat

dasthe

Kindchild

demthe


diethe

Nussnut

morgentomorrow

gebengive

dürfenmay

mussmust‘that it must be possible that it is allowed that the child gives thesquirrel the nut tomorrow’

b. * dassthat

dasthe

Kindchild

demthe


diethe

Nussnut

gebengive

morgentomorrow

dürfenmay

mussmust

c. * dassthat

dasthe

Kindchild

demthe


diethe

Nussnut

gebengive

dürfenmay

morgentomorrow

mussmust


This section deals with embedded clauses that are introduced by a complemen-tizer and with embedded interrogative clauses. The Germanic languages varywith respect to the verb placement in these subordinate clauses and with respectto the question whether the embedded clauses are V2 or not.

2.5.1 Embedded clauses introduced by a complementizer

As was already mentioned Afrikaans, Dutch, German are SOV languages andthis is shown in embedded clauses that are introduced by a complementizer. (31)is an example:


2 Phenomena

(31) IchI

weiß,know

dassthat

AickeAicke

dasthe

Buchbook

heutetoday

gelesenread

hat.has

‘I know that Aicke read the book today.’

English, beeing an SVO non-V2 language allows for SVO order only.

(32) I know that Kim has read the book yesterday. (English)

Interestingly, Danish, also an SVO languauge, allows both SVO order (33) andV2 order (34) in clauses preceeded by a complementizer:

(33) JegI

ved,know

atthat

GertGert

ikkenot

harhas

læstread

bogenbook.DEF

i dag.today

(Danish)

‘I know that Gert did not read the book today.’

(34) a. JegI

ved,know

atthat

i dagtoday

harhas

GertGert

ikkenot

læstread

bogen.book.DEF

(Danish)

b. JegI

ved,know

atthat

bogenbook.DEF

harhas

GertGert

ikkenot

læstread

i dag.today

The example in (33) includes the negation in order to show that we indeed dealwith the SVO order here. Without the negation it is not clear whether non-V2clauses are allowed in clauses that are introduced by a complementizer since(35a) has the finite verb in second position. With the negation present, it is clearthat we have a V2 clause if the negation follows the finite verb and that we donot have a V2 clause if the finite verb follows the negation as in (33) and henceis in third position.

(35) a. atthat

GertGert

harhas

læstread

bogenbook.DEF

(V2 or SVO)

b. atthat

GertGert

harhas

ikkenot

læstread

bogenbook.DEF

(V2)

For complementizerless sentences the V2 order is the only one that is possible:

(36) a. GertGert

harhas

ikkenot

læstread

bogenbook.DEF

(V2)

b. * GertGert

ikkenot

harhas

læstread

bogenbook.DEF

(SVO)

Yiddish and Icelandic are SVO languages as well. The clauses that are com-bined with a complementizer are V2:



(37) a. IkhI

meynthink

azthat

haynttoday

hothas

MaxMax

geleyentread

dosthe

bukh.8

book(Yiddish)

‘I think that Max read the book today.’b. Ikh

Imeynthink

azthat

dosthe

bukhbook

hothas

MaxMax

geleyent.read

(38) Engumno.one.DAT

dattfell

íto

hug,mind

aðthat

vertworth

væriwas

aðto

reynatry

tilPREP

aðto

kynnastknow

honum.9

him(Icelandic)

‘It didn’t occur to anyone that it was worth trying to get to know him.’

2.5.2 Interrogative clauses

TheOV languages form subordinated interrogative clauses by preposing a phrasecontaining an interrogative pronoun10 from an otherwise SOV clause. (39) showsa German example:

(39) a. IchI

weiß,know

werwho

heutetoday

dasthe

Buchbook

gelesenread

hat.has

(German)

‘I know who read the book today.’b. Ich

Iweiß,know

waswhat

AickeAicke

heutetoday

gelesenread

hat.has

‘I know what Aicke has read today.’

Since languages like German allow for scrambling, senteces like those in (39)could just be due to the permutation of arguments of a head. However, the gen-eralization about these w-clauses is that an arbitrary w-element can be fronted.(40) gives an example from German that involves a nonlocal dependency:

(40) IchI

weißknow

nicht,not

[überabout

welcheswhich

Thema]𝑖topic

sieshe

versprochenpromised

hat,has

[[einena

Vortragtalk

_𝑖]to

zuhold

halten]. (German)

‘I do not know about which topic she promised to give a talk.’8Diesing (1990: 58)9Maling (1990: 75)10Most interrogative pronouns start withw in German andwh in English. Phrases containing aninterrogative pronoun are called w-phrases or wh-phrases, respectively. Interrogative clausesare sometimes called w-clauses or wh-clauses.


2 Phenomena

Here, the phrase über welches Thema ‘about which topic’ is an argument of Vor-trag, which is embedded in the VP containing zu halten ‘to hold’, which is in turnembedded under versprochen hat ‘promised has’. The generalization about inter-rogative clauses is that an interrogative clause consists of a interrogative phrase(über welches Thema ‘about which topic’) and a clause in which this interroga-tive phrase is missing somewhere (er versprochen hat, einen Vortrag zu halten ‘hepromised to give a talk’).

In German the order of the other constituents is free as in assertive mainclauses and embedded clauses with a complementizer that were discussed ear-lier.

(41) a. IchI

weiß,know

waswhat

keinernobody

diesemthis


gebengive

würde.would

(German)

‘I know what nobody would give this man.’b. Ich

Iweiß,know

waswhat

diesemthis


keinernobody

gebengive

würde.would

In Danish and English the interrogative clauses consist of an interrogativephrase and an SVO clause in which it is missing:

(42) a. GertGert

harhas

givetgiven

hamhim

bogen.book.DEF

(Danish)

‘Gert gave him the book.’b. Jeg

Ived,know

hvad𝑖what

[GertGert

harhas

givetgiven

hamhim

_𝑖].

‘I know what Gert gave him.’c. Jeg

Ived,know

hvem𝑖

who[GertGert

harhas

givetgiven

_𝑖 bogen].book.DEF

‘I know who Gert has given the book.’

(42a) shows the clause with SVO order and (42b) is an example with the sec-ondary object as interrogative ponoun and (42c) is an example with the primaryobject as interrogative pronoun. The position that the respective objects have innon-interrogative clauses like (42a) is marked with _𝑖 .

Yiddish is special in that it has a V2 order in interrogative clauses as well(Diesing 1990: Sections 4.1, 4.2): interrogatives consist of a interrogative phrasethat is extracted from a V2 clause:


2.6 The use of expletives to mark the clause type

(43) Iryou

veystknow

efshermaybe

[avuwhere

dothere

voyntlives

RoznblatRoznblat

derthe

goldshmid]?11

goldsmith‘Do you perhaps know where Roznblat the goldsmith lives?’

So the variation is w-phrase + SOV, w-phrase + SVO, and w-phrase + V2.

2.6 The use of expletives to mark the clause type

The Germanic languages use constituent order to code the clause type: V2 mainclauses can be assertions or questions, depending on the content of the preverbalmaterial and intonation. Similarly embedded interrogative clauses consist of aw-phrase and an SVO, SOV, or V2 clause. The fronting of a constituent in a V2clause comes with certain information structural effects: something is the topicor the focus of an utterance. For embedded sentences it is important for somelanguages that the structure is transparent that is that we have the w + SVO or w+ V2 order. There are situations in which it is inappropriate to front an elementand in such situations the Germanic languages use expletives, that is, pronounsthat do not contribute semantically, to maintain a certain order.

German uses the expletive es to fill the position in front of the finite verb, ifno other constituent is to be fronted.

(44) a. Dreithree

Reiterriders

rittenrode

zumto.the

Torgate

hinaus.out

(German)

‘Three riders rode out of the gate.’b. Es

EXPLrittenrode

dreithree

Reiterriders

zumto.the

Torgate

hinaus.out

Danish uses the expletive to make it clear that an extraction of a consituenttook place (Müller & Ørsnes 2011: 169):12

(45) a. Politietpolice.DEF

vedknows

ikke,not

hvemwho

derEXPL

havdehas

placeretplaced

bomben.13

bomb.DEF(Danish)

‘The police does not know who placed the bomb.’

11Diesing (1990: 65). Quoted from Olsvanger, Royte Pomerantsn, 1949.12Examples marked with DK are extracted from KorpusDK, a corpus of 56 Million words docu-menting contemporary Danish (http://ordnet.dk/korpusdk).

13DK


http://ordnet.dk/korpusdk

2 Phenomena

b. * Politietpolice.DEF

vedknows

ikke,not

hvemwho

havdehas

placeretplaced

bomben.bomb.DEF

Without the expletive one would have a pattern like the one in (45b). In (45b) wehave the normal SVO order and it is not obvious to the hearer that the patternconsists of an extracted element (the subject) and an SVO clause from which itis missing. This is more transparent if an expletive is inserted into the subjectposition as in (45a). (46) shows this using the analysis that will be suggestedin Chapter 8: (46a) shows the hypothetical structure that would result if oneassumed that the subject hvem ‘who’ is extracted. So-called string-vacuous move-ment would result: the subject is moved to a place right next to it. In (46b) on theother hand the subject position is taken by the expletive and hence it is clear thatthe embedded sentence has a special structure. There is an overt marker for thehearer or reader of the sentence marking it as an embedded interrogative clause.

(46) a. * [hvem𝑖

who[_𝑖 havde

hasplaceretplaced

bomben]]bomb.DEF

(Danish)

b. [hvem𝑖

who[derEXPL

havdehas

_𝑖 placeretplaced

bomben]]bomb.DEF

Similarly Yiddish uses an expletive in embedded interrogatives (w + V2) ifthere is no other element that is information structurally appropriate for thepreverbal position. (47) shows examples from Prince (1989: 403–404):

(47) a. ikhI

hobhave

ziher

gefregtasked

verwho

esEXPL

izis

beserbetter

farfor

irher

(Yiddish)

‘I have asked her who is better for her.’b. ikh

Ihobhave

imhim

gefregtasked

vemenwhom

esEXPL

kenenknow

aleall

dayneyour

khaverimfriends

‘I asked him whom all your friends know.’

(47a) is an example involving an interrogative pronoun that is the subject and(47b) is an example in which the preverbal position is not filled by an argumentof kenen ‘know’ but by an expletive. The subject ale dayne khaverim ‘all yourfriends’ stays behind and the object vemen ‘whom’ is extracted since it is theinterrogative pronoun.

2.7 Verbal Complexes in OV languages

The OV languages have a verbal complex, or more general, a predicate complex,since adjectives take part in complex formation as well. (48) gives a German


2.8 Obligatoriness of subjects, case of subjects, and passives

example by Haider (1986b: 110; 1991: 128):

(48) weilbecause

esit.ACC

ihrher.DAT

jemandsomebody.NOM

zuto

lesenread

versprochenpromised

hathas

(German)

‘because somebody promised her to read it’

The arguments of the respective verbs can be mixed with arguments of otherverbs. In the example above the es ‘it’ is not adjacent to its verb lesen ‘to read’,neither is ihm ‘him’ adjacent to versprochen ‘promised’ nor jemand to hat ‘has’.In a more “well-behaved” ordering the object of zu lesen ‘to read’ is adjacent tothe verb:

(49) weilbecause

jemandsomebody

ihrher

dasthe

Buchbook

zuto

lesenread

versprochenpromised

hathas

‘because somebody promised her to read the book’

The ordering in (49) would allow for an analysis in which das Buch zu lesen formsa VP which is treated as an argument of versprochen ‘promised’. However, this isnot a viable analysis for (48) if one assumes that phrases have to be continuous.

One explanation of orders like the one in (48) is that the verbs form a unitthat behaves like a simplex verb. As with the ditransitive verb geben ‘to give’ allpermutations of the arguments of the verbs are possible in principle. So zu lesenversprochen hat forms a complex in both (48) and (49) and all permutations ofthe three arguments are permitted by the grammar.

VO languages like English andDanish do not allow permutations of argumentsthat belong to different verbs. In VO languages governing verbs always embedVPs. The following example indicates the structure:

(50) because somebody [will [promise him [to read the book]]]

2.8 Obligatoriness of subjects, case of subjects, andpassives

SVO languages like English and Danish require a subject, while OV languageslike German allow for subjectless constructions.

(51) a. Ihmhim.DAT

grautdreads

vorbefore

derthe

Prüfung.exam

(German)

‘He dreads the exam.’


2 Phenomena

b. Heutetoday

wirdis

nichtnot

gearbeitet.worked

‘There is no working today.’

According to the tests that were used for languages like Icelandic (for instancethe possibility to omit a subject in infinitival constructions), the dative object ihm‘him’ in (51a) is not a subject. There is no nominal argument at all in (51b). As wewill discuss in Section 7, Icelandic has dative subjects (Zaenen et al. 1985), whichmakes it the most exciting language to study among the Germanic languages.We will see that a uniform analysis of case assignment is possible (Yip, Maling& Jackendoff 1987), although there is some variety in the inflectional systems ofthe Germanic languages.

As is shown in (51b), German allows for so-called impersonal passives. Imper-sonal passives are a special kind of passives in which no element gets promotedto subject. SVO languages like English, Danish do not allow subjectless construc-tions. English therefore does not allow impersonal passives at all as (52b) shows:

(52) a. weilbecause

nochstill

gearbeitetworked

wirdis

(German)

‘because there is still working there’b. * because (it) was worked (English)

Interestingly, Danish found a way to fulfill the subject requirement and at thesame time have impersonal passives: Danish simply inserts an expletive pronouninto the subject position:

(53) a. fordibecause

derEXPL

bliveris

arbejdetworked

(Danish)

‘because there is working there’b. fordi

becausederEXPL

arbejdeswork.PASS

‘because there is working there’

German does not allow for an expletive subject:

(54) * weilbecause

esit

nochstill

gearbeitetworked

wirdis

(German)

‘because there is still working there’

It is possible to have an expletive pronoun in front of the finite verb as in (55), butthis is a positional expletive whose purpose it is to mark the V2 sentence type.


2.9 Summary

(55) EsEXPL

wirdis

nochstill

gearbeitet.worked

(German)

‘There is still working there.’

The expletive is not an argument of any verb. The purely positional character ofthis expletive is shown by the fact that it does not appear in verb last sentenceslike (54).

2.9 Summary

This chapter provided an overview of the phenomena that are covered in thisbook. Of course we will look at everything in much more detail in the chaptersto come. Let’s start and get our hands dirty.

Comprehension questions

• What is the characteristics of a V2 language?

• If a language has many sentences with subject, verb, object order, doyou know that the language is a V2 language?


3 Phrase structure grammars and Xtheory

This chapter introduces phase structure grammars (PSGs), which play an impor-tant role in many theories that were developed since Chomsky (1957). The phrasestructure grammar developed in this chapter will be the basis for more compli-cated phenomena covered in the chapters to come. This chapter deals mainlywith German and English, which is sufficient for the introduction of the formalapparatus of phrase structure grammars. The result of this chapter is a phrasestructure grammar that is similar to X grammars of the style that was developedin the late 1970ies and the early 1980ies (Jackendoff 1977). The structures arguedfor in this chapter will also play a role in later chapters, but the lexical items willbe much richer: they will contain valence information playing a crucial role inlicensing syntactic structure.

Much time is spent on the structure of noun phrases. The main insights on thesyntax of German noun phrases can be carried over to other Germanic languages.Later chapters will deal with the differences of the Germanic languages withrespect to clause level syntax.

This chapter heavily draws from Müller (2020: Chapter 2), which is an up-dated translation of Müller (2013a: Chapter 2). Knowledge about basic conceptslike part of speech and constituency tests is presupposed. Readers who feel theneed to refresh their knowledge in these areas are referred to Chapter 1 of theGrammatical Theory textbooks.

3.1 Symbols and rewrite rules

Words can be assigned to a particular part of speech on the basis of their in-flectional properties and syntactic distribution. Thus, weil ‘because’ in (1) is aconjunction, whereas das ‘the’ and dem ‘the’ are articles and therefore classedas determiners. Furthermore, Buch ‘book’ and Kind ‘child’ are nouns and gibt‘gives’ is a verb.

3 Phrase structure grammars and 𝑋 Theory

(1) weilbecause

erhe

dasthe

Buchbook

demthe

Kindchild

gibtgives

‘because he gives the child the book’

Using the constituency tests introduced inMüller (2020: Chapter 1), one can showthat individual words as well as the strings das Buch ‘the book’ and dem Kind ‘thechild’, form constituents. These get then assigned certain symbols. Since nounsform an important part of the phrases das Buch and dem Kind, these are referredto as noun phrases or NPs, for short. The pronoun er ‘he’ can occur in the samepositions as full NPs and can therefore also be assigned to the category NP.

Phrase structure grammars come with rules specifying which symbols are as-signed to certain kinds of words and how these are combined to create morecomplex units. A simple phrase structure grammar which can be used to analyze(1) is given in (2):1,2

(2) NP → Det NS → NP NP NP V

NP → erDet → dasDet → dem

N → BuchN → KindV → gibt

We can therefore interpret a rule such as NP → Det N as meaning that a nounphrase, that is, something which is assigned the symbol NP, can consist of adeterminer (Det) and a noun (N).

We can analyze the sentence in (1) using the grammar in (2) in the followingway: first, we take the first word in the sentence and check if there is a rulein which this word occurs on the right-hand side of the rule. If this is the case,then we replace the word with the symbol on the left-hand side of the rule. Thishappens in lines 2–4, 6–7 and 9 of the derivation in (3). For instance, in line 2 eris replaced by NP. If there are two or more symbols which occur together on theright-hand side of a rule, then all these words are replaced with the symbol onthe left. This happens in lines 5, 8 and 10. For instance, in line 5 and 8, Det andN are rewritten as NP.

1I ignore the conjunction weil ‘because’ for now. Since the exact analysis of German verb-first and verb-second clauses requires a number of additional assumptions, we will restrictourselves to verb-final clauses in this chapter.

2The rule NP → er may seem odd. We could assume the rule PersPron → er instead but thenwould have to posit a further rule which would specify that personal pronouns can replace fullNPs: NP → PersPron. The rule in (2) combines the two aforementioned rules and states thater ‘he’ can occur in positions where noun phrases can.



(3) words and symbols rules that are applied

1 er das Buch dem Kind gibt2 NP das Buch dem Kind gibt NP → er3 NP Det Buch dem Kind gibt Det → das4 NP Det N dem Kind gibt N → Buch5 NP NP dem Kind gibt NP → Det N6 NP NP Det Kind gibt Det → dem7 NP NP Det N gibt N → Kind8 NP NP NP gibt NP → Det N9 NP NP NP V V → gibt10 S S → NP NP NP V

In (3), we began with a string of words and it was shown that we can derive thestructure of a sentence by applying the rules of a given phrase structure grammar.We could have applied the same steps in reverse order: starting with the sentencesymbol S, we would have applied the steps 9–1 and arrived at the string of words.Selecting different rules from the grammar for rewriting symbols, we could usethe grammar in (2) to get from S to the string er dem Kind das Buch gibt ‘he thechild the book gives’. We can say that this grammar licenses (or generates) a setof sentences.

The derivation in (3) can also be represented as a tree. This is shown by Fig-ure 3.1. The symbols in the tree are called nodes. We say that S immediately

S

NP

erhe

NP

Det

dasthe

N

Buchbook

NP

Det

demthe

N

Kindchild

V

gibtgives

Figure 3.1: Analysis of er das Buch dem Kind gibt ‘he the book the child gives’

dominates the NP nodes and the V node. The other nodes in the tree are alsodominated, but not immediately dominated, by S. If we want to talk about therelationship between nodes, it is common to use kinship terms. In Figure 3.1, Sis the mother node of the three NP nodes and the V node. The NP node and Vare sisters since they have the same mother node. If a node has two daughters,



then we have a binary branching structure. If there is exactly one daughter, thenwe have a unary branching structure. Two constituents are said to be adjacent ifthey are directly next to each other.

Phrase structure rules are often omitted in linguistic publications. Instead, au-thors opt for tree diagrams or the compact equivalent bracket notation such as(4).

(4) [S [NP er]he

[NP [Det das]the

[N Buch]]book

[NP [Det dem]the

[N Kind]]child

[V gibt]]gives

Nevertheless, it is the grammatical rules which are actually important since theserepresent grammatical knowledge which is independent of specific structures. Inthis way, we can use the grammar in (2) to parse or generate the sentence in (5),which differs from (1) in the order of objects:

(5) [weil]because

erhe.NOM

demthe.DAT

Kindchild

dasthe.ACC

Buchbook

gibtgives

‘because he gives the child the book’

The rules for replacing determiners and nouns are simply applied in a differentorder than in (1). Rather than replacing the first Det with das ‘the’ and the firstnoun with Buch ‘book’, the first Det is replaced with dem ‘the’ and the first nounwith Kind.

At this juncture, I should point out that the grammar in (2) is not the onlypossible grammar for the example sentence in (1). There is an infinite number ofpossible grammars which could be used to analyze these kinds of sentences (seeMüller (2020: Chapter 2) exercise 1). Another possible grammar is given in (6):

(6) NP → Det NV → NP V

NP → erDet → dasDet → dem

N → BuchN → KindV → gibt

This grammar licenses binary branching structures as shown in Figure 3.2 on thefacing page.

Both the grammar in (6) and (2) are too imprecise. If we adopt additionallexical entries for ich ‘I’ and den ‘the’ (accusative) in our grammar, then wewouldincorrectly license the ungrammatical sentences in (7b–d):3

3With the grammar in (6), we also have the additional problem that we cannot determine whenan utterance is complete since the symbol V is used for all combinations of V andNP. Therefore,we can also analyze the sentence in (i) with this grammar:



V

NP

erhe

V

NP

Det

dasthe

N

Buchbook

V

NP

Det

demthe

N

Kindchild

V

gibtgives

Figure 3.2: Analysis of er das Buch dem Kind gibt with a binary branching struc-ture

(7) a. erhe.NOM

dasthe.ACC

Buchbook

demthe.DAT

Kindchild

gibtgives

‘He gives the book to the child.’b. * ich

I.NOMdasthe.ACC

Buchbook

demthe.DAT

Kindchild

gibtgives

c. * erhe.NOM

dasthe.ACC

Buchbook

denthe.ACC

Kindchild

gibtgives

d. * erhe.NOM

denthe.M

Buchbook(N)

demthe

Kindchild

gibtgives

In (7b), subject-verb agreement has been violated, in other words: ich ‘I’ and gibt‘gives’ do not fit together. (7c) is ungrammatical because the case requirements of

(i) a. * derthe.NOM

Delphindolphin

erwartetexpects

b. * desthe.GEN

Kindeschild.GEN

erhe.NOM

dasthe.ACC

Buchbook

demthe.DAT

Kindchild

gibtgives

The number of arguments required by a verb must be somehow represented in the grammar.In Chapter 4, we will see exactly how the selection of arguments by a verb (valence) is capturedin HPSG.



the verb have not been satisfied: gibt ‘gives’ requires a dative object. Finally, (7d)is ungrammatical because there is a lack of agreement between the determinerand the noun. It is not possible to combine den ‘the’, which is masculine andbears accusative case, and Buch ‘book’ because Buch is neuter gender. For thisreason, the gender properties of these two elements are not the same and theelements can therefore not be combined.

In the following, we will consider how we would have to change our grammarto stop it from licensing the sentences in (7b–d). If we want to capture subject-verb agreement, then we have to cover the following six cases in German, as theverb has to agree with the subject in both person (1, 2, 3) and number (sg, pl):

(8) a. IchI

schlafe.sleep

(1, sg)

b. Duyou

schläfst.sleep

(2, sg)

c. Er/sie/eshe/she/it

schläft.sleeps

(3, sg)

d. Wirwe

schlafen.sleep

(1, pl)

e. Ihryou

schlaft.sleep

(2, pl)

f. Siethey

schlafen.sleep

(3, pl)

It is possible to capture these relations with grammatical rules by increasing thenumber of symbols we use. Instead of the rule S → NP NP NP V, we can use thefollowing:

(9) S → NP_1_sg NP NP V_1_sgS → NP_2_sg NP NP V_2_sgS → NP_3_sg NP NP V_3_sgS → NP_1_pl NP NP V_1_plS → NP_2_pl NP NP V_2_plS → NP_3_pl NP NP V_3_pl

This would mean that we need six different symbols for noun phrases and verbsrespectively, as well as six rules rather than one.

In order to account for case assignment by the verb, we can incorporate caseinformation into the symbols in an analogous way. We would then get rules suchas the following:



(10) S → NP_1_sg_nom NP_dat NP_acc V_1_sg_nom_dat_accS → NP_2_sg_nom NP_dat NP_acc V_2_sg_nom_dat_accS → NP_3_sg_nom NP_dat NP_acc V_3_sg_nom_dat_accS → NP_1_pl_nom NP_dat NP_acc V_1_pl_nom_dat_accS → NP_2_pl_nom NP_dat NP_acc V_2_pl_nom_dat_accS → NP_3_pl_nom NP_dat NP_acc V_3_pl_nom_dat_acc

Since it is necessary to differentiate between noun phrases in four cases, we havea total of six symbols for NPs in the nominative and three symbols for NPs withother cases. Since verbs have to match the NPs, that is, we have to differentiatebetween verbs which select three arguments and those selecting only one or two(11), we have to increase the number of symbols we assume for verbs.

(11) a. AickeAicke

schläft.sleeps

‘Aicke is sleeping.’b. * Aicke

Aickeschläftsleeps

dasthe

Buch.book

c. AickeAicke

kenntknows

dasthe

Buch.book

‘Aicke knows the book.’d. * Aicke

Aickekennt.knows

In the rules above, the information about the number of arguments required bya verb is included in the marking ‘nom_dat_acc’.

In order to capture the determiner-noun agreement in (12), we have to incor-porate information about gender (fem, mas, neu), number (sg, pl), case (nom, gen,dat, acc) and the inflectional classes (strong, weak)4.

(12) a. derthe.M

Mann,man(M)

diethe.F

Frau,woman(F)

dasthe.N

Buchbook(N)

(gender)

b. dasthe

Buch,book.SG

diethe

Bücherbooks.PL

(number)

c. desthe.GEN

Buches,book.GEN

demthe.DAT

Buchbook

(case)

d. eina

Beamter,civil.servant

derthe

Beamtecivil.servant

(inflectional class)

4These are inflectional classes for adjectives which are also relevant for some nouns such asBeamter ‘civil servant’, Verwandter ‘relative’, Gesandter ‘envoy’.



Instead of the rule NP → Det N, we will have to use rules such as those in (13):5

(13) NP_3_sg_nom → Det_fem_sg_nom N_fem_sg_nomNP_3_sg_nom → Det_mas_sg_nom N_mas_sg_nomNP_3_sg_nom → Det_neu_sg_nom N_neu_sg_nomNP_3_pl_nom → Det_fem_pl_nom N_fem_pl_nomNP_3_pl_nom → Det_mas_pl_nom N_mas_pl_nomNP_3_pl_nom → Det_neu_pl_nom N_neu_pl_nom

NP_3_sg_nom → Det_fem_sg_nom N_fem_sg_nomNP_3_sg_nom → Det_mas_sg_nom N_mas_sg_nomNP_3_sg_nom → Det_neu_sg_nom N_neu_sg_nomNP_3_pl_nom → Det_fem_pl_nom N_fem_pl_nomNP_3_pl_nom → Det_mas_pl_nom N_mas_pl_nomNP_3_pl_nom → Det_neu_pl_nom N_neu_pl_nom

(13) shows the rules for nominative noun phrases. We would need analogousrules for genitive, dative, and accusative. We would then require 24 symbols fordeterminers (3 ∗ 2 ∗ 4), 24 symbols for nouns and 24 rules rather than one. Ifinflection class is taken into account, the number of symbols and the number ofrules doubles.

3.2 Expanding PSG with features

Phrase structure grammars which only use atomic symbols are problematic asthey cannot capture certain generalizations. We as linguists can recognize thatNP_3_sg_nom stands for a noun phrase because it contains the letters NP. How-ever, in formal terms this symbol is just like any other symbol in the grammarand we cannot capture the commonalities of all the symbols used for NPs. Fur-thermore, unstructured symbols do not capture the fact that the rules in (13) allhave something in common. In formal terms, the only thing that the rules havein common is that there is one symbol on the left-hand side of the rule and twoon the right.

We can solve this problem by introducing features which are assigned to cate-gory symbols and therefore allow for the values of such features to be includedin our rules. For example, we can assume the features person, number and casefor the category symbol NP. For determiners and nouns, we would adopt an ad-

5To keep things simple, these rules do not incorporate information regarding the inflectionclass.


3.3 Phrase structure rules for some aspects of German syntax

ditional feature for gender and one for inflectional class. (14) shows two rulesaugmented by the respective values in brackets:6

(14) NP(3,sg,nom) → Det(fem,sg,nom) N(fem,sg,nom)NP(3,sg,nom) → Det(mas,sg,nom) N(mas,sg,nom)

If we were to use variables rather than the values in (14), we would get ruleschemata as the one in (15):

(15) NP(3,Num,Case) → Det(Gen,Num,Case) N(Gen,Num,Case)

The values of the variables here are not important. What is important is thatthey match. For this to work, it is important that the values are ordered; that is,in the category of a determiner, the gender is always first, number second andso on. The value of the person feature (the first position in the NP(3,Num,Case))is fixed at ‘3’ by the rule. These kind of restrictions on the values can, of course,be determined in the lexicon:

(16) NP(3,sg,nom) → esDet(mas,sg,gen) → des

The rules in (10) can be collapsed into a single schema as in (17):

(17) S → NP(Per1,Num1,nom)NP(Per2,Num2,dat)NP(Per3,Num3,acc)V(Per1,Num1,ditransitive)

The identification of Per1 and Num1 on the verb and on the subject ensures thatthere is subject-verb agreement. For the other NPs, the values of these featuresare irrelevant. The case of these NPs is explicitly determined.

3.3 Phrase structure rules for some aspects of Germansyntax

Whereas determining the direct constituents of a sentence is relative easy, sincewe can very much rely on the movement test due to the somewhat flexible orderof constituents in German, it is more difficult to identify the parts of the noun

6In the following chapters, attribute value structures will be used. In these structures, we alwayshave pairs of a feature name and a feature value. In such a setting, the order of values is notimportant, since every value is uniquely identified by the corresponding feature name. Sincewe do not have a feature name in schemata like (13), the order of the values is important.



phrase. This is the problem we will focus on in this section. To help motivateassumptions about X syntax to be discussed in Section 3.4, we will also discussprepositional phrases.

3.3.1 Noun phrases

Up to now, we have assumed a relatively simple structure for noun phrases: ourrules state that a noun phrase consists of a determiner and a noun. Noun phrasescan have a distinctly more complex structure than (18a). This is shown by thefollowing examples in (18):

(18) a. eina

Buchbook

b. eina

Buch,book

dasthat

wirwe

kennenknow

c. eina

Buchbook

ausfrom

JapanJapan

d. einan

interessantesinteresting

Buchbook

e. eina

Buchbook

ausfrom

Japan,Japan

dasthat

wirwe

kennenknow

f. einan


Buchbook

ausfrom

JapanJapan

g. einan


Buch,book

dasthat

wirwe

kennenknow

h. einan


Buchbook

ausfrom

Japan,Japan

dasthat

wirwe

kennenknow

As well as determiners and nouns, noun phrases can also contain adjectives,prepositional phrases and relative clauses. The additional elements in (18) are ad-juncts. They restrict the set of objects which the noun phrase refers to. Whereas(18a) refers to a being which has the property of being a book, the referent of(18b) must also have the property of being known to us.

Our previous rules for noun phrases simply combined a noun and a determinerand can therefore only be used to analyze (18a). The questionwe are facing now ishow we can modify this rule or which additional rules we would have to assume



in order to analyze the other noun phrases in (18). In addition to rule (19a), onecould propose a rule such as the one in (19b).7,8

(19) a. NP → Det Nb. NP → Det A N

However, this rule would still not allow us to analyze noun phrases such as (20):

(20) alleall

weiterenfurther

schlagkräftigenstrong

Argumentearguments

‘all other strong arguments’

In order to be able to analyze (20), we require a rule such as (21):

(21) NP → Det A A N

It is always possible to increase the number of adjectives in a noun phrase andsetting an upper limit for adjectives would be entirely arbitrary. Even if we optfor the following abbreviation, there are still problems:

(22) NP → Det A* N

The asterisk in (22) stands for any number of iterations. Therefore, (22) encom-passes rules with no adjectives as well as those with one, two or more.

The problem is that according to the rule in (22) adjectives and nouns do notform a constituent and we can therefore not explain why coordination is stillpossible in (23):

(23) alleall

[[großenbig

Seeelefanten]elephant.seals

undand

[grauengrey

Eichhörnchen]]squirrels

‘all the big elephant seals and grey squirrels’

If we assume that coordination involves the combination of two or more wordstrings with the same syntactic properties, then we would have to assume thatthe adjective and noun form a unit.

The following rules capture the noun phrases with adjectives discussed thusfar:

(24) a. NP → Det Nb. N → A N

7See Eisenberg (2004: 238) for the assumption of flat structures in noun phrases.8There are, of course, other features such as gender and number, which should be part of all therules discussed in this section. I have omitted these in the following for ease of exposition.



c. N → N

These rules state the following: a noun phrase consists of a determiner and anominal element (N). This nominal element can consist of an adjective and anominal element (24b), or just a noun (24c). Since N is also on the right-handside of the rule in (24b), we can apply this rule multiple times and thereforeaccount for noun phrases with multiple adjectives such as (20). Figure 3.3 showsthe structure of a noun phrase without an adjective and that of a noun phrasewith one or two adjectives. The adjective grau ‘grey’ restricts the set of referents

NP

Det

eina

N

N


NP

Det

eina

N

A

grauesgrey

N

N


NP

Det

eina

N

A

großesbig

N

A

grauesgrey

N

N


Figure 3.3: Noun phrases with differing numbers of adjectives

for the noun phrase. If we assume an additional adjective such as groß ‘big’, thenit only refers to those squirrels who are grey as well as big. These kinds of nounphrases can be used in contexts such as the following:

(25) A: Alleall

grauengrey

Eichhörnchensquirrels

sindare

groß.big

‘All grey squirrels are big.’B: Nein,

noichI

habehave

eina

kleinessmall

grauesgrey


gesehen.seen

‘No, I saw a small grey squirrel.’

We observe that this discourse can be continued with Aber alle kleinen grauenEichhörnchen sind krank ‘but all small grey squirrels are ill’ and a correspondinganswer. The possibility to have even more adjectives in noun phrases such as ein



kleines graues Eichhörnchen ‘a small grey squirrel’ is accounted for in our rulesystem in (24). In the rule (24b), N occurs on the left as well as the right-handside of the rule. This kind of rule is referred to as recursive.

We have now developed a nifty little grammar that can be used to analyzenoun phrases containing adjectival modifiers. As a result, the combination of anadjective and noun is given constituent status. One may wonder at this point ifit would not make sense to also assume that determiners and adjectives form aconstituent, as we also have the following kind of noun phrases:

(26) diesethese

schlauensmart

undand

diesethese

neugierigencurious


Here, we are dealing with a different structure, however. Two full NPs have beenconjoined and part of the first conjunct has been deleted.

(27) diesethese

schlauensmart


undand

diesethese

neugierigencurious


One can find similar phenomena at the sentence and even word level:

(28) a. dassthat

ConnyConny

demthe

Kindchild

dasthe

Buchbook

gibtgives

undand

AickeAicke

derthe

Frauwoman

diethe

Schallplatterecord

gibtgives

‘that Conny gives the book to the child and Aicke the record to thewoman’

b. be-PRFX

undand

ent-ladenPRFX-load

‘load and unload’

Thus far, we have discussed howwe can ideally integrate adjectives into our rulesfor the structure of noun phrases. Other adjuncts such as prepositional phrasesor relative clauses can be combined with N in an analogous way to adjectives:

(29) a. N → N PPb. N → N relative clause

With these rules and those in (24), it is possible – assuming the correspondingrules for PPs and relative clauses – to analyze all the examples in (18).

(24c) states that it is possible for N to consist of a single noun. A further im-portant rule has not yet been discussed: we need another rule to combine nounssuch as Vater ‘father’, Sohn ‘son’ or Bild ‘picture’, so-called relational nouns, with



their arguments. Examples of these can be found in (30a–b). (30c) is an exampleof a nominalization of a verb with its argument:

(30) a. derthe

Vaterfather

vonof

PeterPeter

‘Peter’s father’b. das

theBildpicture

vomof.the

GleimtunnelGleimtunnel

‘the picture of the Gleimtunnel’c. das

theKommencoming

derof.the

Installateurinplumber

‘the plumber’s visit’

The rule that we need to analyze (30a,b) is given in (31):

(31) N → N PP

Figure 3.4 shows two structures with PP-arguments. The tree on the right alsocontains an additional PP-adjunct, which is licensed by the rule in (29a).

NP

Det

dasthe

N

N

Bildpicture

PP

vom Gleimtunnelof.the Gleimtunnel

NP

Det

dasthe

N

N

N

Bildpicture

PP


PP

im Gropiusbauin.the Gropiusbau

Figure 3.4: Combination of a noun with PP complement vom Gleimtunnel to theright with an adjunct PP

In addition to the previously discussed NP structures, there are other struc-tures where the determiner or noun is missing. Nouns can be omitted via ellipsis.(32) gives an example of noun phrases, where a noun that does not require acomplement has been omitted. The examples in (33) show NPs in which onlyone determiner and complement of the noun has been realized, but not the nounitself. The underscore marks the position where the noun would normally occur.



(32) a. einan


_

‘an interesting one’b. ein

aneuesnew


_

‘a new interesting one’c. ein

aninteressantesinteresting

_ ausfrom

JapanJapan

‘an interesting one from Japan’d. ein

aninteressantesinteresting

_, dasthat

wirwe

kennenknow

‘an interesting one that we know’

(33) a. (Nein,no

nichtnot

derthe

Vaterfather

vonof

Klaus),Klaus

derthe

_ vonof

PeterPeter

warwas

gemeint.meant

‘No, it wasn’t the father of Klaus, but rather the one of Peter that wasmeant.’

b. (Nein,no

nichtnot

dasthe

Bildpicture

vonof

derthe

Stadtautobahn),motorway

dasthe

_ vomof.the

GleimtunnelGleimtunnel

warwas

beeindruckend.impressive

‘No, it wasn’t the picture of the motorway, but rather the one of theGleimtunnel that was impressive.’

c. (Nein,no

nichtnot

dasthe

Kommencoming

desof.the

Tischlers),carpenter

dasthe

_ derof.the

Installateurinplumber

istis

wichtig.important

‘No, it isn’t the visit of the carpenter, but rather the visit of theplumber that is important.’

In English, the pronoun one must often be used in the corresponding position,9

but in German the noun is simply omitted. In phrase structure grammars, thiscan be described by a so-called epsilon production. These rules replace a symbolwith nothing (34a). The rule in (34b) is an equivalent variant which is responsiblefor the term epsilon production:

9See Fillmore et al. (2012: Section 4.12) for English examples without the pronoun one.



(34) a. N →b. N → 𝜖

The corresponding trees are shown in Figure 3.5. Going back to boxes, the rules

NP

Det

einan

N

A


N

N

_

NP

Det

dasthe

N

N

_

PP


Figure 3.5: Noun phrases without an overt head

in (34) correspond to empty boxes with the same labels as the boxes of ordinarynouns. As we have considered previously, the actual content of the boxes isunimportant when considering the question of where we can incorporate them.In this way, the noun phrases in (18) can occur in the same sentences. The emptynoun box also behaves like one with a genuine noun. If we do not open the emptybox, we will not be able to ascertain the difference to a filled box.

It is not only possible to omit the noun from noun phrases, but the determinercan also remain unrealized in certain contexts. (35) shows noun phrases in plural:

(35) a. Bücherbooks

b. Bücher,books

diethat

wirwe

kennenknow

c. interessanteinteresting

Bücherbooks

d. interessanteinteresting

Bücher,books

diethat

wirwe

kennenknow

The determiner can also be omitted in singular if the noun denotes a mass noun:



(36) a. Getreidegrain

b. Getreide,grain

dasthat

geradejust

gemahlenground

wurdewas

‘grain that has just been ground’c. frisches

freshGetreidegrain

d. frischesfresh

Getreide,grain

dasthat

geradejust

gemahlenground

wurdewas

‘fresh grain that has just been ground’

Finally, both the determiner and the noun can be omitted:

(37) a. IchI

leseread

interessante.interesting

‘I read interesting ones.’b. Dort

theredrübenover

stehtstands

frisches,fresh

dasthat

geradejust

gemahlenground

wurde.was

‘Over there is some fresh (grain) that has just been ground.’

Figure 3.6 shows the corresponding trees.

NP

Det

_

N

N

Bücherbooks

NP

Det

_

N

A

interessanteinteresting

N

N

_

Figure 3.6: Noun phrases without overt determiner

It is necessary to add two further comments to the rules were developed up tothis point: up to now, I have always spoken of adjectives. However, it is possible



to have very complex adjective phrases in pre-nominal position. These can beadjectives with complements (38a,b) or adjectival participles (38c,d):

(38) a. derthe

seinerhis.DAT

Frauwife

treuefaithful

Mannman

‘the man faithful to his wife’b. der

theaufon

seinenhis.ACC

Sohnson

stolzeproud

Mannman

‘the man proud of his son’c. der

theseinehis.ACC

Frauwoman

liebendeloving

Mannman

‘the man who loves his wife’d. der

thevonby

seinerhis.DAT

Frauwife

geliebteloved

Mannman

‘the man loved by his wife’

Taking this into account, the rule (24b) has to be modified in the following way:

(39) N → AP N

An adjective phrase (AP) can consist of an NP and an adjective, a PP and anadjective or just an adjective:

(40) a. AP → NP Ab. AP → PP Ac. AP → A

There are two imperfections resulting from the rules were developed thus far.These are the rules for adjectives or nouns without complements in (40c) as wellas (24c) – repeated here as (41):

(41) N → N

If we apply these rules, then we will generate unary branching subtrees, that istrees with a mother that only has one daughter. See Figure 3.6 for an example ofthis. If we maintain the parallel to the boxes, this would mean that there is a boxwhich contains another box which is the one with the relevant content.

In principle, nothing stops us from placing this information directly into thelarger box. Instead of the rules in (42), we will simply use the rules in (43):

(42) a. A → kluge



b. N → Mann

(43) a. AP → klugeb. N → Mann

(43a) states that kluge ‘smart’ has the same properties as a full adjective phrase,in particular that it cannot be combined with a complement. This is parallel tothe categorization of the pronoun er ‘he’ as an NP in the grammars (2) and (6).

Assigning N to nouns which do not require a complement has the advantagethat we do not have to explain why the analysis in (44b) is possible as well as(44a) despite there not being any difference in meaning.

(44) a. [NP einigesome

[N klugesmart

[N [N [N Frauenwomen

] undand

[N [N Männermen

]]]]]]

b. [NP einigesome

[N klugesmart

[N [N [N Frauenwomen

] undand

[N Männermen

]]]]]

In (44a), two nouns have projected to N and have then been joined by coordi-nation. The result of coordination of two constituents of the same category isalways a new constituent with that category. In the case of (44a), this is also N.This constituent is then combined with the adjective and the determiner. In (44b),the nouns themselves have been coordinated. The result of this is always anotherconstituent which has the same category as its parts. In this case, this would beN. This N becomes N and is then combined with the adjective. If nouns whichdo not require complements were categorized as N rather than N, we would nothave the problem of spurious ambiguities. The structure in (45) shows the onlypossible analysis.

(45) [NP einigesome

[N klugesmart

[N [N Frauenwomen

] undand

[N Männermen

]]]]

3.3.2 Prepositional phrases

Compared to the syntax of noun phrases, the syntax of prepositional phrases(PPs) is relatively straightforward. PPs normally consist of a preposition and anoun phrase whose case is determined by that preposition. We can capture thiswith the following rule:

(46) PP → P NP



This rule must, of course, also contain information about the case of the NP. Ihave omitted this for ease of exposition as I did with the NP-rules and AP-rulesabove.

The Duden grammar (Eisenberg et al. 2005: §1300) offers examples such asthose in (47), which show that certain prepositional phrases serve to further de-fine the semantic contribution of the preposition by indicating some measure-ment, for example:

(47) a. [[Einenone

Schritt]step

vorbefore

demthe

Abgrund]abyss

bliebremained

erhe

stehen.stand

‘He stopped one step in front of the abyss.’b. [[Kurz]

shortlynachafter

demthe

Start]take.off

fielfell

diethe

Klimaanlageair.conditioning

aus.out

‘Shortly after take off, the air conditioning stopped working.’c. [[Schräg]

diagonallyhinterbehind

derthe

Scheune]barn

istis

eina

Weiher.pond

‘There is a pond diagonally across from the barn.’d. [[Mitten]

middleimin.the

Urwald]jungle

stießenstumbled

diethe

Forscherresearchers

aufon

einenan

altenold

Tempel.temple‘In the middle of the jungle, the researches came across an old temple.’

To analyze the sentences in (47a,b), one could propose the following rules in (48):

(48) a. PP → NP PPb. PP → AP PP

These rules combine a PP with an indication of measurement. The resultingconstituent is another PP. It is possible to use these rules to analyze prepositionalphrases in (47a,b), but it unfortunately also allows us to analyze those in (49):

(49) a. * [PP einenone

Schrittstep

[PP kurzshortly

[PP vorbefore

demthe

Abgrund]]]abyss

b. * [PP kurzshortly

[PP einenone

Schrittstep

[PP vorbefore

demthe

Abgrund]]]abyss

Both rules in (48) were used to analyze the examples in (49). Since the symbol PPoccurs on both the left and right-hand side of the rules, we can apply the rulesin any order and as many times as we like.



We can avoid this undesired side-effect by reformulating the previously as-sumed rules:

(50) a. PP → NP Pb. PP → AP Pc. PP → Pd. P → P NP

Rule (46) becomes (50d). The rule in (50c) states that a PP can consist of P. Fig-ure 3.7 shows the analysis of (51) using (50c) and (50d) as well as the analysis ofan example with an adjective in the first position following the rules in (50b) and(50d):

(51) vorbefore

demthe

Abgrundabyss

‘in front of the abyss’

PP

P

P

vorbefore

NP

dem Abgrundthe abyss

PP

AP

kurzshortly

P

P

vorbefore

NP

dem Abgrundthe abyss

Figure 3.7: Prepositional phrases with and without measurement

At this point, the attentive reader is probablywonderingwhy there is no emptymeasurement phrase in the left figure of Figure 3.7, which one might expect inanalogy to the empty determiner in Figure 3.6. The reason for the empty deter-miner in Figure 3.6 is that the entire noun phrase without the determiner has ameaning similar to those with a determiner. The meaning normally contributedby the visible determiner has to somehow be incorporated in the structure of thenoun phrase. This can be done by the empty determiner with an appropriatelyspecified meaning contribution.

Unlike determiner-less NPs, prepositional phrases without an indication ofdegree or measurement do not lack any meaning component for composition. It



is therefore not necessary to assume an empty indication of measurement, whichsomehow contributes to the meaning of the entire PP. Hence, the rule in (50c)states that a prepositional phrase consists of P, that is, a combination of P andNP.

3.4 X theory

If we look again at the rules that were formulated in the previous section, wesee that heads are always combined with their complements to form a new con-stituent (52a,b), which can then be combined with further constituents (52c,d):

(52) a. N → N PPb. P → P NPc. NP → Det Nd. PP → NP P

Grammarians working on English noticed that parallel structures can be used forphrases which have adjectives or verbs as their head. I discuss adjective phrasesat this point and postpone the discussion of verb phrases to Chapter 4, since theassumptions regarding the structure of clauses in both German and English de-viate from X theory as it is commonly assumed today. As in German, certainadjectives in English can take complements with the important restriction thatadjective phrases with complements cannot realize these pre-nominally in En-glish. (53) gives some examples of adjective phrases:

(53) a. Kim and Sandy are proud.b. Kim and Sandy are very proud.c. Kim and Sandy are proud of their child.d. Kim and Sandy are very proud of their child.

Unlike prepositional phrases, complements of adjectives are normally optional.proud can be used with or without a PP. The degree expression very is also op-tional.

The rules which we need for this analysis are given in (54), with the corre-


3.4 𝑋 Theory

sponding structures in Figure 3.8.

(54) a. AP → Ab. AP → AdvP Ac. A → A PPd. A → A

AP

A

A

proud

AP

AdvP

very

A

A

proud

AP

A

A

proud

PP

of their child

AP

AdvP

very

A

A

proud

PP

of their child

Figure 3.8: English adjective phrases

As was shown in Section 3.2, it is possible to generalize over very specificphrase structure rules and thereby arrive at more general rules. In this way, prop-erties such as person, number and gender are no longer encoded in the categorysymbols, but rather only simple symbols such as NP, Det and N are used. It isonly necessary to specify something about the values of a feature if it is relevantin the context of a given rule. We can take this abstraction a step further: insteadof using explicit category symbols such as N, V, P and A for lexical categoriesand NP, VP, PP and AP for phrasal categories, one can simply use a variable forthe word class in question and speak of X and XP.

This form of abstraction can be found in so-called X theory (or X-bar theory,the term bar refers to the line above the symbol), which was developed by Chom-sky (1970) and refined by Jackendoff (1977). This form of abstract rules plays animportant role in many different theories. For example: Government & Bind-ing (Chomsky 1981), Generalized Phrase Structure Grammar (Gazdar et al. 1985;Uszkoreit 1987) and Lexical Functional Grammar (Bresnan 1982; Bresnan et al.2016). In HPSG, the theory assumed in this book, X theory also plays a role, butnot all restrictions of the X schema have been adopted.

(55) shows a possible instantiation of X rules, where the category X has beenused in place of N, as well as examples of word strings which can be derived bythese rules:



(55) X rule with specific categories example strings

X → specifier X N → DET N the [picture of Paris]

X → X adjunct N → N REL_CLAUSE [picture of Paris][that everybody knows]

X → adjunct X N → A N beautiful [picture of Paris]

X → X complement∗ N → N P picture [of Paris]

Any word class can replace X (e.g., V, A or P). The X without the bar stands fora lexical item in the above rules. If one wants to make the bar level explicit, thenit is possible to write X0. Just as with the rule in (15), where we did not specifythe case value of the determiner or the noun but rather simply required that thevalues on the right-hand side of the rule match, the rules in (55) require that theword class of an element on the right-hand side of the rule (X or X) matches that

of the element on the left-hand side of the rule (X or X).A lexical element can be combined with all its complements. The ‘*’ in the last

rule stands for an unlimited amount of repetitions of the symbol it follows. Aspecial case is zerofold occurrence of complements. There is no PP complementof Bild ‘picture’ present in das Bild ‘the picture’ and thus N becomes N. The resultof the combination of a lexical element with its complements is a new projectionlevel of X: the projection level 1, which is marked by a bar. X can then be com-bined with adjuncts. These can occur to the left or right of X. The result of thiscombination is still X, that is the projection level is not changed by combiningit with an adjunct. Maximal projections are marked by two bars. One can alsowrite XP for a projection of X with two bars. An XP consists of a specifier and X.Depending on one’s theoretical assumptions, subjects of sentences (Haider 1995;1997a; Berman 2003: Section 3.2.2) and determiners in NPs (Chomsky 1970: 210)are specifiers. Furthermore, degree modifiers (Chomsky 1970: 210) in adjectivephrases and measurement indicators in prepositional phrases are also countedas specifiers.

Non-head positions can only host maximal projections and therefore comple-ments, adjuncts and specifiers always have two bars. As already mentionedabove, HPSG does not stick to the X theory. For example, arguments maybewords or intermediate, non-complete phrases (see Chapter 5 about the verbalcomplex in Germanic SOV languages). Figure 3.9 on the facing page gives anoverview of the minimal and maximal structure of phrases.

Some categories do not have a specifier or have the option of having one. Ad-juncts are optional and therefore not all structures have to contain an X with anadjunct daughter. In addition to the branching shown in the right-hand figure,


3.4 𝑋 Theory

XP

X

X

XP

specifier X

adjunct X

complement X

Figure 3.9: Minimal and maximal structure of phrases

adjuncts to XP and head-adjuncts are sometimes possible. There is only a singlerule in (55) for cases in which a head precedes the complements, however an or-der in which the complement precedes the head is of course also possible. Thisis shown in Figure 3.9.

Figure 3.10 on the next page shows the analysis of the NP structures das Bild‘the picture’ and das schöne Bild von Paris ‘the beautiful picture of Paris’. The NPstructures in Figure 3.10 and the tree for proud in Figure 3.8 show examples ofminimally populated structures. The left tree in Figure 3.10 is also an example ofa structure without an adjunct. The right-hand structure in Figure 3.10 is an ex-ample for the maximally populated structure: specifier, adjunct, and complementare present.

The analysis given in Figure 3.10 assumes that all non-heads in a rule arephrases. One therefore has to assume that there is a determiner phrase evenif the determiner is not combined with other elements. The unary branching ofdeterminers is not elegant but it is consistent.10 The unary branchings for theNP Paris in Figure 3.10 may also seem somewhat odd, but they actually becomemore plausible when one considers more complex noun phrases:

(56) a. dasthe

ParisParis

derof.the

dreißigerthirty

Jahreyears

‘30’s Paris’b. die

theMariaMaria

ausfrom

HamburgHamburg

‘Maria from Hamburg’

10For an alternative version of X theory which does not assume elaborate structure for deter-miners see Muysken (1982).



NP

DetP

Det

Det

dasthe

N

N

Bildpicture

NP

DetP

Det

Det

dasthe

N

AP

A

A

schönebeautiful

N

N

Bildpicture

PP

P

P

vonof

NP

N

N

ParisParis

Figure 3.10: X analysis of das Bild ‘the picture’ and das schöne Bild von Paris ‘thebeautiful picture of Paris’

Unary projections are somewhat inelegant but this should not concern us toomuch here, as we have already seen in the discussion of the lexical entries in (43)that unary branching nodes can be avoided for the most part and that it is indeeddesirable to avoid such structures. Otherwise, one gets spurious ambiguities. Inthe following chapters, I show how HPSG (light) can analyze determiners andnoun, adjective, and verb phrases without assuming unary rules. So instead ofassuming the structures in Figure 3.10, the much simpler ones in Figure 3.11 willbe used.


3.4 𝑋 Theory

NP

Det

dasthe

N

Bildpicture

NP

Det

dasthe

N

AP

schönebeautiful

N

N

Bildpicture

PP

P

vonof

NP

ParisParis

Figure 3.11: HPSG analysis of das Bild ‘the picture’ and das schöne Bild von Paris‘the beautiful picture of Paris’


1. Why are phrase structure grammars that use only atomic categoriesinadequate for the description of natural languages?

2. Assuming the grammar in (6), state which steps (replacing symbols)one has to take to get to the symbol V in the sentence (57).

(57) erhe

dasthe

Buchbook

demthe

Kindchild

gibtgives

‘He gives the book to the child.’

Your answer should resemble the analysis in (3).



Exercises

This chapter is in large part identical to Chapter 2 of Müller (2020). Sincethe focus of this book is different from the textbook about grammaticaltheory, I decided to provide a different set of exercises here. Those whoare interested in doingmore exercises may consult the grammatical theorytext book in addition. It is published by Language Science Press and henceopen access, that is, it is freely available.

1. Draw trees for the following phrases. You may use the symbol NP forproper names and N for nouns not requiring complements (as in Fig-ure 3.11).

(58) a. eineone

Stundehour

vorbefore

derthe

Ankunftarrival

desof.the

Zugestrain

‘one hour before the arrival of the train’b. kurz

shortlynachafter

derthe

Ankunftarrival

inin

ParisParis

‘shortly after the arrival in Paris’c. das

thiseina

Liedsong

singendesinging

Kindchild

ausfrom

demthe

AllgäuAllgäu

‘the child from the Allgäu singing a song’

2. Use the online version of SWI-Prologa to test your grammar using acomputer. Details regarding the notation can be found in the EnglishWikipedia entry for Definite Clause Grammar (DCG).b

ahttps://swish.swi-prolog.org/, 2020-06-07.bhttps://en.wikipedia.org/wiki/Definite_clause_grammar, 2020-06-07.


https://swish.swi-prolog.org/

https://en.wikipedia.org/wiki/Definite_clause_grammar

3.4 X theory

Further reading

The expansion of phrase structure grammars to include features was pro-posed as early as 1963 by Harman (1963).

The phrase structure grammar for noun phrases discussed in this chap-ter covers a large part of the syntax of noun phrases but cannot explaincertain NP structures. Furthermore, it has the problem, which exercise 3of Müller (2020: Chapter 2) is designed to show. A discussion of these phe-nomena and a solution in the framework of HPSG can be found in Netter(1998) and Kiss (2005).

The discussion of the integration of semantic information into phrasestructure grammars was very short. A detailed discussion of predicatelogic and its integration into phrase structure grammars – as well as adiscussion of quantifier scope – can be found in Blackburn & Bos (2005).


4 Valence, argument order and adjunctplacement

This chapter deals with the representation of valency information and sketchesthe basic structures that are assumed for SVO and SOV languages. I provide anaccount for scrambling in those languages that allow for it and discuss the fixedvs. free position of adjuncts.

4.1 Valence representations

The word sequences in (1) were already discussed in footnote 3 on page 36.

(1) a. * derthe.NOM

Delphindolphin

erwartetexpects

b. * desthe.GEN

Kindeschild.GEN

derthe.NOM

Delphindolphin

denthe.ACC

Ballball

demthe.DAT

Kindchild

gibtgives

The problem is that there are too few (1a) or too many NPs (1b) present. Theconcept that is needed here is valence: like in chemistry it is assumed that headshave a certain potential to enter into stable relations with other material (Tes-nière 2015: 239). For example, the verb erwarten ‘to expect’ requires an NP inthe nominative and one in the accusative. geben ‘to give’ is the prototypical di-transitive verb: it can be combined with an NP in the nominative, an NP in thedative and an NP in the accusative, but as (1b) shows, a genitive object could notbe integrated into a sentence.

The NPs in the examples in (2) are arguments of the respective verbs:

(2) a. [dass]that

derthe.NOM

Delphindolphin

denden.ACC

Menschenhuman

erwartetexpects

‘that the dolphin expects the human’b. [dass]

thatderthe.NOM

Delphindolphin

demthe.DAT

Kindchild

denthe.ACC

Ballball

gibtgives

‘that the dolphin gives the child the ball’

4 Valence, argument order and adjunct placement

Most syntactic arguments also fill a so-called semantic role in the semantic rep-resentation of the head. For example, the dolphin is the giver, the child is therecipient, and the ball is the item given. Tesnière (2015: Chapter 48) suggestedusing the analogy of dramas for the explanation of valence: if we imagine thescene of giving, what has to happen on stage to call an event that is acted outa giving event? There have to be the three participants, a giver, a recipient andsomething that is given. Without these participants, we do not have a propergiving event.

In addition to elements like the NPs in the examples above, which are calledarguments, there are also so-called adjuncts. schnell ‘quickly’ and quickly areexamples for adjuncts:

(3) a. [dass]that

derthe.NOM

Delphindolphin

demthe.DAT

Kindchild

schnellquickly

denthe.ACC

Ballball

gibtgives

‘that the dolphin gives the child the ball quickly’b. that the dolphin gives the child the ball quickly

The adverbials provide additional information about the giving event, but theydo not fill a semantic role.

Tomake things complicated not all arguments have to be realized in a sentence.The ditransitive verb geben can be realized with any subset of its arguments,provided the context fills in the missing information.

(4) a. Sieshe

gibtgives

Geld.money

‘She gives money.’b. Sie

shegibtgives

denthe

Armen.poor

‘She gives to the poor.’c. Sie

shegibt.gives

d. Gib!give

In the case of (4a), a certain charity setting could have been established and onecan either donate food or money or contribute some voluntary work. In such asituation, (4a) is perfectly fine. The transfered object in (4b) is probably money.A possible context for (4c) and (4d) is the card game skat where the person whodeals out is rotating among the players. (4d) is an imperative. Even subjects can



be dropped in imperatives since the referent of the subject is obvious: it is theaddressee of the utterance.

The examples in (4) show that the arguments of geben ‘to give’ may be omit-ted. This is not the case for the accusative object of erwarten ‘to expect’: it isobligatory. So arguments may be optional or obligatory, but adjuncts are alwaysoptional. While the number of argument is limited (by the number of availableslots), the number of adjuncts is not: there can be arbitrarily many adjuncts in aphrase. (5) shows an example with two adjuncts:

(5) [dass]that

derthe.NOM

Delphindolphin

jetztnow

demthe.DAT

Kindchild

schnellquickly

denthe.ACC

Ballball

gibtgives

‘that the dolphin now gives the child the ball quickly’

The analogy to chemistry and dramamay be confusing sinceH2O is a very niceand stable molecule and it is helpful to imagine the parallel combination of a verbwith its two arguments. Figure 4.1 shows H2O and the parallel combination ofa verb with its arguments corresponding to (6a). The problem is that a single Hand an O do not form a stable combination, while (6b) is fine:

(6) a. Kirby helps Sandy.b. Kirby helps.

O

H H

helps

Kirby Sandy

Figure 4.1: Combination of hydrogen and oxygen and the combination of a verbwith its arguments

Of course one can simply assume that there is a version of helps that has a valencedifferent from the two-place valence usually assumed. Here is were the parallelbreaks down since we do not have an oxygen atom with just one open slot forthe hydrogen atom. The drama analogy adds to the confusion since the helpingevent described in (6b) of course involves somebody who is helped. The solutionto this problem is to distinguish between syntactic and semantic valence. Thedrama analogy helps us to find the semantic valence, the chemistry analogy ismore about syntactic valence.

Given that chemistry and drama have their problems, we may go for anotheranalogy: food. Let’s assume you want to prepare a meal with pasta, tofu and a



tomato sauce. For the tomato sauce you also need some onions. You put all theingredients onto a shopping list and go to the shop. Once in the shop you realizethat they run out of tofu. Your meal will work without tofu. Tofu is optional.Fortunately, the shop has plenty of pasta. You may choose between the differenttypes and select the pasta type and brand you prefer. Some onions, tomatoes andyour done. Wait, next to the cashier there are these gummy bears. OK, you takesome of these as well although you did not want to and they have nothing to dowith your meal and your shopping list. The gummy bears are the adjuncts.

Back to linguistics: there are two ways of ensuring that arguments are realizedtogether with their heads. The first one uses techniques that were introduced inChapter 3. If one uses flat phrase structure rules, one can make sure that certainarguments appear together with certain heads. A schema similar to the one in(7) was discussed as (17) on page 41.

(7) S → NP[nom] NP[dat] NP[acc] V[ditransitive]

Such schemata have been used in Generalized Phrase Structure Grammar (Gaz-dar, Klein, Pullum & Sag 1985; Uszkoreit 1987), but they were abandoned laterin favor of lexicalist models, that is, models assuming that information aboutarguments of a head is encoded in the lexical description of the head ratherthan in phrase structure rules (Jacobson 1987; Müller 2016: Section 5.5; Müller& Wechsler 2014a). Reasons for abandoning the phrasal approach of GPSG wereproblems with so-called partial verb phrase frontings (Nerbonne 1986; Johnson1986) and with accounting for interactions with morphology (Müller 2016: Sec-tion 5.5.1).1

In lexical approaches, the valence of a head is represented in its lexical entryin the form of a list with descriptions of the elements that belong to the head’svalence. (8) provides some prototypical examples:

1Starting with influential work by Adele Goldberg (1995) in the framework of ConstructionGrammar, the phrasal approaches had a revival (Goldberg & Jackendoff 2004). Phrasal ap-proaches are wide-spread and also assumed in other frameworks Haugereid 2007; 2009; Culi-cover & Jackendoff 2005; Alsina 1996; Christie 2010; Asudeh et al. 2008; 2013. The problemsthat lead to the abandonment of GPSG are ignored in the literature and newly introduced onesare not properly addressed. See Müller 2006; 2010; 2013c; Müller & Wechsler 2014a,b; Müller2017; 2018; 2019; 2020; 2021c for some discussion. Note that there are also lexical variants ofConstruction Grammar. Sag, Boas & Kay (2012), introducing Sign-Based Construction Gram-mar, explicitly argue for a lexical view citing some of the references just given. I am workingin the framework of Constructional HPSG (Sag 1997), which is also underlying the proposalssketched in this book.



(8) a. schläft ‘sleeps’: ⟨ NP[nom] ⟩b. kennt ‘knows’: ⟨ NP[nom], NP[acc] ⟩c. hilft ‘helps’: ⟨ NP[nom], NP[dat] ⟩d. gibt ‘gives’: ⟨ NP[nom], NP[dat], NP[acc] ⟩e. wartet ‘waits’:

⟨NP[nom], PP[auf ]

⟩The elements in such lists come with a fixed order. The order corresponds tothe order of the elements in English and to the so-called unmarked order in Ger-man, that is, for ditransitive verbs the order is usually nom, dat, acc (see Höhle(1982) for comments on the unmarked order). This fixed order is needed for es-tablishing the link between syntax and semantics. This will be briefly discussedin Section 4.8.

Given such a valence representation for a verb like kennen ‘know’, one canassume a schema that combines an element from the valence list with the respec-tive head and passes all unsaturated elements on to the result of the combination.The first step of the analysis of (9) is provided in Figure 4.2.2

(9) [dass]that

niemandnobody.NOM

ihnhim.ACC

kenntknows

‘that nobody knows him’

V ⟨ NP[nom] ⟩

NP[acc]

ihnhim

V ⟨ NP[nom], NP[acc] ⟩

kenntknows

Figure 4.2: Analysis of ihn kennt ‘him knows’, valence information is representedin a list

The lexical item for kennt ‘knows’ has a valence description containing two NPs.In a first step kennt is combined with its accusative object. The resulting phraseihn kennt ‘him knows’ is something whose most importent constituent is a verb.

2Note that this sounds as if there were an order in which things have to be combined. This isnot the case. HPSG grammars are sets of constraints that can be applied in any order. It is forexplanatory purposes only that analyses are explained in a bottom up fashion throughout thebook.



Therefore it has V as its category label. Certain important properties of linguis-tic objects are called head features. Part of speech is one of these properties. It isassumed that all head features are passed up from the head in the tree automati-cally.

The element that is not yet combined with kennt ‘knows’ is the NP[nom]. It isstill represented in the valence list of ihn kennt ‘him knows’. Figure 4.3 shows thenext step combining ihn kennt with the subject niemand ‘nobody’. The result is

V ⟨⟩

NP[nom]

niemandnobody

V ⟨ NP[nom] ⟩

NP[acc]

ihnhim


kenntknows

Figure 4.3: Analysis of (dass) niemand ihn kennt ‘that nobody knows him’

a linguistic object of category verb with the empty list as valence representation.As will be shown shortly, the schema that licenses structures like the V ⟨⟩ and

V ⟨ NP[nom] ⟩ in Figure 4.3 is a more abstract version of the rule in (6) on page 36.It is probably helpful to return to our meal-shopping analogy. Assume we are

using an app to organize our shopping lists. For our current meal we need pastaand tomatoes. They are listed in the app in a certain order (tomatoes, pasta) andthere are little images attached to the products. Once we found somethingmatch-ing the pasta, we remove the pasta from the list and the remaining list containsan icon reminding us of the tomatoes. Once we have those, we remove themfrom the list and since nothing is left on the list, we pay. Linguistic structuresare similar: we start with a verb selecting two NPs, we combine it with one NPand then with the second one. The result is a complete structure, something withan empty valence list.

There are various ways to deal with optional arguments. The most simple oneis to assume further lexical items selecting fewer arguments. For the example in(4c) one would assume the valence representation in (10a) and for sentences withwarten ‘to wait’ without prepositional object, one would assume (10b) in additionto the representations in (8):


4.2 Scrambling

(10) a. gibt ‘gives’: ⟨ NP[nom] ⟩b. wartet ‘waits’: ⟨ NP[nom] ⟩

4.2 Scrambling

As we already saw in the data discussion in Chapter 2.3, some languages allowfor scrambling of arguments. For those languages one can assume that heads cancombine with any of its arguments not necessarily beginning with the last oneas it was the case in the analysis in Figure 4.3. Figure 4.4 shows the analysis of(11).

(11) [dass]that

ihnhim.ACC

niemandnobody.NOM

kenntknows


V ⟨⟩

NP[acc]

ihnhim

V ⟨ NP[acc] ⟩

NP[nom]

niemandnobody


kenntknows

Figure 4.4: Analysis of (dass) ihn niemand kennt ‘that nobody knows him’, lan-guages that allow for scrambling permit the saturation of argumentsin any order

Rather than combining the verb with the accusative argument (the object) first,it is combined with the nominative (the subject) and the accusative (the object)is added in a later step.

4.3 SVO: Languages with fixed SV order and valencefeatures

The last section demonstrated how verb-final sentences in German can be ana-lyzed. Of course it is easy to imagine how this extends to VSO languages: The



head is initial and combines with the first element in the valence list first andthen with all the other elements. However, nothing has been said about the SVOlanguages so far. In languages like Danish, English, and so on all objects arerealized after the verb as in (12), it is just the subject that preceedes the verb.

(12) Kim gave Sandy the book.

The verb together with its objects forms a unit in a certain sense: it can be fronted(13a). It can be selected by dominating verbs (13b), it can be coordinated (13c), andit is the place where adjuncts attach to (13d–e).

(13) a. John promised to read the book and [read the book], he will.b. He will [read the book].c. Kim [[sold the car] and [bought a bicycle]].d. He often [reads the book].e. … [often [read the book] slowly], he will.

This can be modeled adequately by assuming two valence lists: one for the com-plements (COMPS short for COMPLEMENTS) and one for the subject. The list forthe subject is called SPECIFIER list (SPR).3 The specifier list plays a role both inthe analysis of sentences and in the analysis of noun phrases. Nouns select theirdeterminer via SPR and all their other arguments via COMPS. Figure 4.5 on thefacing page shows the analysis of the sentence (14) using the features SPR andCOMPS.

(14) Nobody knows him.

The COMPS list of knows contains a description of the accusative object and the ac-cusative him is combined in a first step with knows. In addition to the accusativeobject knows selects for a subject. This selection is passed on to the mother node,

3There are various versions of HPSG: Pollard & Sag (1987: Chapter 3.2) assumed that all argu-ments of a head are represented in one list. This list was called SUBCAT list. Borsley (1987) ar-gued that one should use several valence features (SUBJ, SPR, and COMPS) and this was adoptedin Pollard & Sag (1994: Chapter 9): subjects of verbs were selected via SUBJ and determinersvia SPR. Sag, Wasow & Bender (2003: Chapter 4.3) assume that both subjects and determinersare selected via SPR, which is what is assumed in the grammars developed here too. Sag (2012:Section 3.3) presents a versions of HPSG called Sign-Based Construction Grammar (SBCG)which assumes one valence list for all arguments a it was common in 1987. This return to anabandoned approach came without any argumentation. Hence, I do not adopt this variant ofHPSG but stick to the separation of subjects and other arguments to SPR and COMPS. I will notuse SUBJ as a valence feature, but it will be introduced in the analysis of verbal complexes inChapter 5.


4.3 SVO: Languages with fixed SV order and valence features

V[SPR ⟨⟩, COMPS ⟨⟩]

NP[nom]

nobody

V[SPR ⟨ NP[nom] ⟩, COMPS ⟨⟩]

V[SPR ⟨ NP[nom] ⟩,COMPS ⟨ NP[acc] ⟩]

knows

NP[acc]

him

= S

= VP

Figure 4.5: Analysis of the SVO order with two separate valence features

the VP. Hence, the SPR value of knows him is identical to the SPR value of knows.The VP knows him selects for a nominative NP. This NP is realized as nobody inFigure 4.5. The result of the combination of knows him with nobody is nobodyknows him, which is complete: It has both an empty SPR list and an empty COMPSlist. The two rules that are responsible for the combinations in Figure 4.5 arecalled the Specifier-Head Schema and the Head-Complement Schema. I use VPas abbreviation for something with a verbal head and an empty COMPS list andat least one element in the SPR list and S as abbreviation for something with averbal head and empty lists for both the SPR and the COMPS value.

In Section 4.2 it was explained how scrambling can be accounted for: the rulesthat combine heads with their arguments can take the arguments from the list inany order. For languages with stricter constituent order requirements the rulesare stricter: the arguments have to be taken off the list consistently from thebeginning or from the end. So for English and Danish one starts at the beginningof the list and for head-final languages without scrambling one starts at the endof the list. Figure 4.6 shows the analysis of a sentence with a ditransitive verb.The accusative object is the first element in the COMPS list and it is combined withthe verb first. The result of the combination is a verbal projection that has thePP[to] as the sole element in the COMPS list. It is combined with an appropriatePP in the next step resulting in a verbal projection that has an empty COMPS list(a VP).

The analysis of our first German example in Figure 4.3 did not use a namefor the valence list. So the question is: How does the analysis of German relateto the analysis of English using SPR and COMPS. A lot of researchers from vari-ous frameworks argued that it is not useful to distinguish the subjects of finite




NP[nom]

Kim

V[SPR ⟨ NP[nom] ⟩, COMPS ⟨⟩]

V[SPR ⟨ NP[nom] ⟩,COMPS ⟨ PP[to] ⟩]

V[SPR ⟨ NP[nom] ⟩,COMPS ⟨ NP[acc], PP[to] ⟩]

gave

NP[acc]

a book

PP[to]

to Sandy

Figure 4.6: Analysis of the SVO order with two separate valence features and twoelements in COMPS

verbs from other arguments. All the tests that have been used to show that sub-jects in English differ from complements do not apply to the arguments of finiteverbs in German. Hence, researchers like Pollard (1996), Haider (1993), Eisenberg(1994: 376), and Kiss (1995) argued for so-called subject as complement analy-ses.Figure 4.7 on the next page shows the adapted analysis of (9) – repeated hereas (15):

(15) [dass]that

niemandnobody.NOM

ihnhim.ACC

kenntknows


The difference between German and English is that German contains all argu-ments in the COMPS list of the finite verb and no arguments in the SPR list. Sincethe elements in the COMPS list can be combined with the head in any order, it isexplained why all permutations of arguments are possible. Specifiers are realizedto the left of their head. This is the same for German and English. For Germanthis is not relevant in the verbal domain, but the Specifier-Head Schema, whichis introduced shortly, is used in the analysis of noun phrases.

Throughout the remainder of this book I use the abbreviations in Table 4.1 onthe facing page.


4.4 Immediate dominance schemata


NP[nom]

niemandnobody

V[SPR ⟨ ⟩, COMPS ⟨ NP[nom] ⟩ ]

NP[acc]

ihnhim

V[SPR ⟨ ⟩,COMPS ⟨ NP[nom], NP[acc]⟩]

kenntknows

= S

= V′

Figure 4.7: The analysis of a German sentence with SPR and COMPS list

Table 4.1: Abbreviations for S, VP, and V′ and NP, N′

S = V[SPR ⟨⟩, COMPS ⟨⟩]VP = V[SPR ⟨ NP[nom] ⟩, COMPS ⟨⟩]V′ = all other V projections apart from verbal complexesNP = N[SPR ⟨⟩, COMPS ⟨⟩]N′ = N[SPR ⟨ DET ⟩, COMPS ⟨⟩]


In Section 4.1 I already mentioned that the non-terminal nodes in a tree, that is,the nodes that are not the leaves of the tree, are licensed by schemata similarto those introduced in Chapter 3.2 and 3.4. In fact, the schemata are even moreabstract than X schemata since they do not make any statements about linearorder of the daughters. The two schemata discussed in this section are sketchedhere as (16):

(16) Specifier-Head Schema and Head-Complement Schema (preliminary)H[SPR 1 ] → H[SPR 1 ⊕ ⟨ 2 ⟩, COMPS ⟨⟩] 2

H[COMPS 1 ] → H[COMPS ⟨ 2 ⟩ ⊕ 1 ] 2

Syntactic rules are usually called schemata since they are rather abstract. Thedetails about such schemata are given in more formal HPSG literature likeMüller(2013b: Chapter 4) orAB2021a, but Figure 4.8 on the next page and Figure 4.11 onpage 76 provide the respective tree representations. The H stands for head. Theterm head daughter is used for the daughter that either is the head of a phrase or



H[SPR 1 ]

2 H[SPR 1 ⊕ ⟨ 2 ⟩,COMPS ⟨⟩]

Figure 4.8: Sketch of the Specifier-Head Schema (preliminary)

contains the head of the phrase (e.g., the verb in a sentence or the noun in a nounphrase). append (⊕) is a relation that concatenates two lists. For instance theconcatenation of ⟨ a ⟩ and ⟨ b ⟩ is ⟨ a, b ⟩. The concatenation of the empty list ⟨⟩with another list yields the latter list. To give some examples that are of relevancein this chapter consider the list ⟨ NP[nom], NP[dat], NP[acc ] ⟩. append can beused to append two lists resulting in our list in the following ways:

(17) a. ⟨⟩ ⊕ ⟨ NP[nom], NP[dat], NP[acc] ⟩ = ⟨ NP[nom], NP[dat], NP[acc] ⟩b. ⟨ NP[nom] ⟩ ⊕ ⟨ NP[dat], NP[acc] ⟩ = ⟨ NP[nom], NP[dat], NP[acc] ⟩c. ⟨ NP[nom], NP[dat] ⟩ ⊕ ⟨ NP[acc] ⟩ = ⟨ NP[nom], NP[dat], NP[acc] ⟩d. ⟨ NP[nom], NP[dat], NP[acc] ⟩ ⊕ ⟨⟩ = ⟨ NP[nom], NP[dat], NP[acc] ⟩

The schema in Figure 4.8 takes a list apart in such a way that a list with a single-ton element ( ⟨ 2 ⟩ ) and a remaining list ( 1 ) results. Assuming the three-elementlist with nom, dat and acc elements, this would be the case in (17c) and 2 wouldbe NP[acc] and 1 would be ⟨ NP[nom], NP[dat] ⟩. In this book, the SPR list hasat most one element.4 It can be an NP[nom] in the case of verbs in the SVO lan-guages or the determiner, if the head is a noun. If one splits a list with a singletonelement into a list containing one element and a rest, the rest will always be theempty list. Hence, with the lists at the right-hand side of the equations in (18), 1

will be the empty list and 2 will be NP[nom] and Det, respectively.

(18) a. ⟨⟩ ⊕ ⟨ NP[nom] ⟩ = ⟨ NP[nom] ⟩b. ⟨⟩ ⊕ ⟨ DET ⟩ = ⟨ DET ⟩

For a schema like the one in Figure 4.8 to apply, the descriptions of the daugh-ters have to match the actual daughters. For instance sleeps is compatible withthe right daughter: it has an NP[nom] in its SPR list. When sleeps is realized asa daughter of the schema in Figure 4.8, 2 is instantiated as NP[nom]. Therefore

4But see Müller & Ørsnes (2013b) for an analysis of object shift in Danish assuming multipleelements in the SPR list.



the left daughter has to be compatible with an NP[nom]. It can be realized as asimple pronoun like she or a complex NP like the brown squirrel. Two analysesare shown in Figure 4.9. The 1 in Figure 4.9 says that whatever is in the SPR list is

V[SPR ⟨⟩,COMPS ⟨⟩]

1 NP[nom]

she

V[SPR ⟨ 1 NP[nom] ⟩,COMPS ⟨⟩]

sleeps

V[SPR ⟨⟩,COMPS ⟨⟩]

1 NP[nom]

the brown squirrel

V[SPR ⟨ 1 NP[nom] ⟩,COMPS ⟨⟩]

sleeps

Figure 4.9: Head-Specifier phrases with a subject and an intransitive verb

identified with whatever is the other element in the tree. I wrote down NP[nom]following the 1 in both the NP node and within the SPR list, but it would havebeen sufficient to mention NP[nom] at one of the two places. The actual numberin the box does not matter. What matters is where the same number appearsin the trees or structures. I usually start with 1 at the top of the tree and useconsecutive number for the following sharings.

Figure 4.12 on p. 77 below shows an example analysis with a ditransitive verbalso involving the Specifier-Head Schema. The specification of the COMPS valueof the head daughter in the Specifier-Head Schema ensures that the verb is com-bined with its complements before the specifier is added.

Apart from its use for the analysis of subject–VP combinations in the SVOlanguages, the Specifier-Head Schema is also used for the analysis of NPs in allthe Germanic languages. Figure 4.10 shows the analysis of the NP the squirrel.squirrel selects for a determiner and the result of combining squirrel with a de-

N[SPR ⟨⟩, COMPS ⟨⟩]

1 Det

the

N[SPR ⟨ 1 ⟩, COMPS ⟨⟩]

squirrel

Figure 4.10: Analysis of the NP the squirrel



terminer is a complete nominal projection, that is, an NP. There are also nounslike picture that take a complement:

(19) a picture of Kim

The combination of picture and its complement of Kim is parallel to the combina-tion of a verb with its object in VO languages with fixed constituent order. Forsuch combinations we need a separate schema: the Head-Complement Schema,which is given in Figure 4.11. The schema splits the COMPS list of a head into an

H[COMPS 1 ]

H[COMPS ⟨ 2 ⟩ ⊕ 1 ] 2

Figure 4.11: Sketch of the Head-Complement Schema (preliminary)

initial list with one element ( 2 ), which is realized as the complement daughterto the right.5 This schema licenses both the combination of gave and the childand the combination of gave the child and a book in Figure 4.12 on the facingpage, which shows the analysis of (20).6

(20) Nobody gave the child a book.

To keep things simple, the Specifier-Head Schema did not mention the COMPSvalue of the mother. The Head-Complement Schema did neither mention the SPRvalue of the head daughter nor the one of the mother. But the respective valuesare important, since something has to be said about these values in structuresthat are licensed by these schemata. If the SPR value in the combination of gaveand the child would not be constrained by the Head-Complement Schema, anvalue would be possible. This includes a SPR list containing two genitive NPsand an accusative NP. Sequences like (21) would be licensed:

(21) * his his him gave the child a book

5In principle daughters are unordered inHPSG as theywere in GPSG (Gazdar et al. 1985). Speciallinearization rules are used to order a head with respect to its siblings in a local tree. So aschema licensing a tree like the one in Figure 4.11 would also license a tree with the daughtersin a different order unless one had linearization rules that rule this out. See Müller (2021a) foran overview of approaches to constituent order in HPSG. Linear precedence rules are discussedin more detail in Section 4.6.

6English nouns and determiners do not inflect for case. However, case ismanifested at pronouns:he (nominative), his (genitive), him (accusative). Hence, verbs in double object constructionsselect for two accusatives rather than for dative and accusative as in German.




1 NP[nom]

nobody

V[SPR ⟨ 1 NP[nom] ⟩, COMPS ⟨⟩]

V[SPR ⟨ NP[nom] ⟩,COMPS ⟨ 2 NP[acc] ⟩]

V[SPR ⟨ NP[nom] ⟩,COMPS ⟨ 3 NP[acc], 2 NP[acc]⟩]

gave

3 NP[acc]

the child

2 NP[acc]

a book

Figure 4.12: Analysis of the sentences with a ditransitive verb

To avoid such unspecified SPR values, the SPR value of the head daughter is iden-tified with the SPR value of the mother node in the schema. This is the 1 in (22b).Similarly, the COMPS value of the mother in Specifier-Head phrases has to bespecified to be identical to the COMPS value of the head daughter ( 2 in (22a)) andhence the empty list.

(22) Specifier-Head Schema and Head-Complement Schema (final)a. H[SPR 1 , COMPS 2 ] → H[SPR 1 ⊕ ⟨ 3 ⟩, COMPS 2 ⟨⟩] 3

b. H[SPR 1 , COMPS 2 ] → H[SPR 1 , COMPS 2 ⊕ ⟨ 3 ⟩] 3

Figure 4.13 shows the final versions of the two schemata.

H[SPR 1 ,COMPS 2 ]

3 H[SPR 1 ⊕ ⟨ 3 ⟩,COMPS 2 ⟨⟩]

H[SPR 1 ,COMPS 2 ]

H[SPR 1 ,COMPS ⟨ 3 ⟩ ⊕ 2 ]]

3

Figure 4.13: Sketch of the Specifier-Head and Head-Complement Schema



4.5 Scrambling and free VO/OV order

Now, in order to analyze languages with free constituent order, we need a moreliberal variant of the schema in Figure 4.11. Figure 4.14 splits the COMPS list ofa head into three parts: a list 1 , a list containing exactly one element ⟨ 3 ⟩ anda third list 2 . The element of the second list is realized as the complement ofthe head. The length of the lists 1 and 2 is not restricted. For our example list

H[COMPS 1 ⊕ 2 ]

3 H[COMPS 1 ⊕ ⟨ 3 ⟩ ⊕ 2 ]

Figure 4.14: Sketch of theHead-Complement Schema for languageswith free con-stituent order

containing a nom, a dat and an acc element, there are the following possibilitiesto split the list:

(23) a. ⟨⟩ ⊕ ⟨ NP[nom] ⟩ ⊕ ⟨ NP[dat], NP[acc] ⟩b. ⟨ NP[nom] ⟩ ⊕ ⟨ NP[dat] ⟩ ⊕ ⟨ NP[acc] ⟩c. ⟨ NP[nom], NP[dat] ⟩ ⊕ ⟨ NP[acc] ⟩ ⊕ ⟨⟩

So 3 in Figure 4.14 would be NP[nom] in (23a), NP[dat] in (23b) and NP[acc] in(23c).

If one restricts 1 to be the empty list, one gets grammars that saturate com-plements from the beginning of the list (VO languages with fixed order like En-glish) and if one restricts 2 to be the empty list, one gets grammars that take thelast element from the COMPS list for combination with a head (this would be anOV languages with fixed order, if such a language would exist). Scrambling lan-guages like German allow any complement to be combined with its head sincethere is neither a restriction on 1 nor one on 2 .

4.6 Linear precedence rules

The abstract schemata are similar to the schemata that we gained by abstractingover simple phrase structure rules in Chapter 3. They are similar to abstractX rules. However, there is an important difference: the elements at the right-hand side of a rule and the daughters in the corresponding treelets in the figures


4.6 Linear precedence rules

visualizing the schemata are not ordered. This means that a schema like the onein (24) can be used to analyze configurations with a preceeding b and with bpreceeding a.

(24) m → a b

As will be shown shortly, this comes handy in situations in which one wants toleave the actual order underspecified.

For the Head Complement Schema discussed above this means that actuallytwo orders can be analyzed: head-daughter before complement and complementbefore head-daughter. Hence the Head-Complement Schema is general enoughto analyze the German and English phrases in (25):

(25) a. demthe

Kindchild

einthe

Buchbook

gibtgives

b. gives the child the book

But such a general schema without restrictions would also allow an analysis for(26b) and (26c):

(26) a. [dass]that

niemandnobody

demthe

Kindchild

eina

Buchbook

vorliestPART.reads

‘that nobody reads a book to the child’b. * [dass]

thatdemthe

Kindchild

niemandnobody

vorliestPART.reads

eina

Buchbook

c. * [dass]that

niemandnobody

vorliestPART.reads

demthe

Kindchild

eina

Buchbook

The structures licensed by the Head-Complement Schema without any restric-tions are shown in the Figures 4.15 and 4.16 on the next page

Now, this problem is easy to fix: what is needed is binary feature specifyingwhether a head is initial or not. The feature is called INITIAL (abbreviated as INI).All head-daughters that are INI+ are always serialized to the left of their com-plement and all those that are INI− are serialized to the right. The linearizationrules are provided in (27):

(27) a. HEAD [INITIAL+] < COMPLEMENTb. COMPLEMENT < HEAD [INITIAL−]

German verbs are specified to be INITIAL−, while English verbs are INITIAL+. Be-cause of this specification and the linearization rules in (27), verbs are always



V[COMPS ⟨ ⟩]

V[COMPS ⟨ 1 ⟩]

2 NP[dat]

dem Kindthe child

V[COMPS ⟨ 2 , 1 ⟩]

3 NP[nom]

niemandnobody

V[COMPS ⟨ 3 , 2 , 1 ⟩]

vorliestPART.reads

1 NP[acc]

ein Bucha book

Figure 4.15: Unwanted analysis using theHead-Complement Schemawithout lin-earization constraints

V[COMPS ⟨ ⟩]

V[COMPS ⟨ 1 ⟩]

V[COMPS ⟨ 2 , 1 ⟩]

3 NP[nom]

niemandnobody

V[COMPS ⟨ 3 , 2 , 1 ⟩]

vorliestPART.reads

2 NP[dat]

dem Kindthe child

1 NP[acc]

ein Bucha book

Figure 4.16: Unwanted analysis using theHead-Complement Schemawithout lin-earization constraints


4.7 Adjuncts

ordered after their complements in German (and other SOV languages) and be-fore their complements in English (and other SVO languages). Of course, thereare sentences in German in which the verb is in first or second position and thereare sentences in the Germanic SVO languages in which the object precedes theverb. These sentences will be covered in Chapter 6.

4.7 Adjuncts

While arguments are selected by their head, adjuncts select the head. The dif-ference between languages like Dutch and German on the one hand and Danishand English on the other hand can be explained by assuming that adjuncts in theformer languages are less picky as far as the element is concerned with whichthey combine. Dutch (28) and German (29) adjuncts can attach to any verbalprojection, while Danish (30) and English (31) require a VP (see also Section 2.4):

(28) a. [dat]that

onmiddellijkpromptly

iedereeneverybody

hetthe

boekbook

leestreads

(Dutch)

‘that everybody reads the book promptly’b. [dat]

thatiedereeneverybody


hetthe

boekbook

leestreads

c. [dat]that

iedereeneverybody

hetthe

boekbook


leestreads

(29) a. [dass]that

sofortpromptly

jedereverybody

dasthe

Buchbook

liestreads

(German)

‘that everybody reads the book promptly’b. [dass]

thatjedereverybody

sofortpromptly

dasthe

Buchbook

liestreads

c. [dass]that

jedereverybody

dasthe

Buchbook

sofortpromptly

liestreads

(30) a. atthat

hvereverybody

læstreads

bogenbook.DEF

strakspromptly

(Danish)

‘that everybody reads the book promptly’b. at

thathvereverybody

strakspromptly

læstreads

bogenbook.DEF

‘that everybody promptly reads the book’



(31) a. that everybody reads the book promptlyb. that everybody promptly reads the book

For the selection of arguments the features SPR and COMPS are used. In parallelthere is a MOD feature that is part of a lexical description of a head of a phrasethat can function as an adjunct (MOD is an abbreviation for modified). The valueof MOD is a description of an appropriate head. Head-adjunct structures are li-cencsed by the schema in Figure 4.17. For instance, attributive adjectives have

H[SPR 1 , COMPS 2 ]

[MOD 3 , SPR ⟨⟩, COMPS ⟨⟩] 3 H[SPR 1 , COMPS 2 ]

Figure 4.17: Sketch of the Head-Adjunct Schema

N as their MOD value, where N is an abbreviation for a nominal projection thathas an empty COMPS list and a SPR list that contains a determiner. (32) shows thelexical item for brown:

(32) Lexical item for brown:PHON ⟨ brown ⟩MOD NSPR ⟨⟩COMPS ⟨⟩

The analysis of the phrase brown squirrel is shown in Figure 4.18. In languages

N

Adj[MOD 2 ]

brown

2 N

squirrel

Figure 4.18: Analysis of the head-adjunct structure brown squirrel

like German in which the adjective agrees with the noun in gender, number, andinflection class, the properties that the nounmust have can be specified inside theMOD value. For instance, kleiner selects a male noun and kleine selects a femaleone:


4.7 Adjuncts

(33) a. eina

kleinerlittle

Hunddog

b. einea

kleinelittle

Katzecat

For German adverbials the value restricts the part of speech of the head tobe verb (or rather verbal since – as (34b) shows – adjectival participles can bemodified as well) and the value of INITIAL to be −.

(34) a. dassthat

esit

oftoften

lachtlaughs

(German)

‘that he/she laughs often’b. des

theoftoften

lachendelaughing

Kindchild

‘the child who laughs often’

The specification of themodified element to be INITIAL− ensures that the adjunctattaches to verbs in final position only (Verb initial sentences are discussed inChapter 6). A linearization rule has to make sure that adverbials are serializedto the left of the verb, that is, somewhere in the Mittelfeld. The MOD value ofEnglish adverbials is simply VP. Without any further restrictions, this allows fora pre- and a post-VP attachment of adjuncts.

• SOV (Dutch, German, …): MOD V[INI−]

• SVO (Danish, English, …): MOD VP

The analysis of (29a) is shown in Figure 4.19 on the following page, the one of(29b) in Figure 4.20 on the next page, and the one of (29c) in Figure 4.21 on page 85.The only difference between the figures is the respective place of attachment ofthe adverb. I marked the parts of the tree that are licensed by the Head-AdjunctSchema by including them into a box. All other nodes in the tree are licensed bythe Head-Complement Schema in Figure 4.11.

The attentive reader will notice that there is a description following the 3 inthe MOD value of the adverbials, while there is no such description in the MODvalues of the English examples that follow. Of course this is purely notationalsince the numbered boxes identify all values with the same numbers, but theconvention behind this is to state the description if it differs from what is givenat other places where the box occurs. In the case of German, the MOD value ofadverbials is just verb without any restrictions regarding valence features. The




Adv[MOD 3 V]

morgentomorrow

3 V[SPR ⟨⟩, COMPS ⟨⟩]

1 NP[nom]

AickeAicke

V[SPR ⟨ ⟩, COMPS ⟨ 1 ⟩ ]

2 NP[acc]

das Buchthe book

V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩]

liestreads

Figure 4.19: Analysis of [dass] morgen jeder das Buch liest ‘that everybody willread the book tomorrow’ with the adjunct attaching above subjectand object


1 NP[nom]

AickeAicke

V[SPR ⟨ ⟩, COMPS ⟨ 1 ⟩ ]

Adv[MOD 3 V]

morgentomorrow

3 V[SPR ⟨ ⟩, COMPS ⟨ 1 ⟩ ]

2 NP[acc]

das Buchthe book

V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩]

liestreads

Figure 4.20: Analysis of [dass] jeder morgen das Buch liest ‘that everybody willread the book tomorrow’ with the adjunct attaching between subjectand object


4.7 Adjuncts


1 NP[nom]

AickeAicke

V[SPR ⟨ ⟩, COMPS ⟨ 1 ⟩ ]

2 NP[acc]

das Buchthe book

V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩]

Adv[MOD 3 V]

morgentomorrow

3 V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩]

liestreads

Figure 4.21: Analysis of [dass] jeder das Buch morgen liest ‘that everybody willread the book tomorrow’ with the adjunct attaching between objectand verb

valence features are given at the modified node (e.g., SPR ⟨⟩, COMPS ⟨ 1 , 2 ⟩ inFigure 4.21), but not in the MOD value. Since English adverbials modify VPs andsince the modified node is a VP, the value of the MOD value is not given in detailin the Figures below, but is just shared with the properties of the modified node.

The Figures 4.22 and 4.23 show the analysis of adjunction with the adverb inpre-VP and post-VP position respectively.

The values of SPR and COMPS in the schema in Figure 4.17 on page 82 havenot been explained so far. First there is the sharing of the SPR and COMPS valuesbetween mother and head-daughter. Whatever element an adjunct attaches to,the valence requirements of the mother are always identical to the valence re-quirement of the head-daughter. Nothing is added, nothing is missing. Adjunctsare additional elements that are not selected for via valence features, hence noth-ing has to be discharged. This can be seen by looking at the German examples inFigures 4.19 to 4.21: morgen ‘tomorrow’ attaches to a nodewith certain valence re-quirements and the dominating node has exactly the same valence requirements.In principle the figures should have little numbered boxes in them indicatingthe identity of the valence requirements of mother and head daughter in head-adjunct combinations. I omitted these so-called structure sharings to keep thingssimple and readable.




1 NP[nom]

Kim

V[SPR ⟨ 1 ⟩, COMPS ⟨ ⟩ ]

Adv[MOD 3 ]

often

3 V[SPR ⟨ 1 ⟩, COMPS ⟨ ⟩ ]

V[SPR ⟨ 1 ⟩,COMPS ⟨ 2 ⟩]

reads

2 NP[acc]

books

Figure 4.22: Analysis of adjuncts in SVO languages: the adjunct is realized left-adjacent to the VP.


1 NP[nom]

Kim

V[SPR ⟨ 1 ⟩, COMPS ⟨ ⟩ ]

3 V[SPR ⟨ 1 ⟩, COMPS ⟨ ⟩ ]

V[SPR ⟨ 1 ⟩,COMPS ⟨ 2 ⟩]

reads

2 NP[acc]

books

Adv[MOD 3 ]

often

Figure 4.23: Analysis of adjuncts in SVO languages: the adjunct is realized right-adjacent to the VP.


4.7 Adjuncts

The adjunct itself has to have empty valence lists, that is, it has to be complete.Without this requirement, sentences like the one in (35) would be licensed:

(35) * Sandy read the book in.

in is a preposition that has an NP[acc] in its COMPS list. If the Head-AdjunctSchema would not specify the COMPS list of the adjunct daughter to be empty, apreposition could function as the adjunct daughter and a structure for ungram-matical sentence like (35) would be licensed by the grammar.

The specifier specification is as important as the specification of the COMPS list.If non-empty SPR lists were allowed, the contrast in (36) could not be explained:

(36) a. dassthat

AickeAicke

einean

Stundehour

liestreads

(German)

‘Aicke is reading for an hour.’b. * dass

thatAickeAicke

Stundehour

liestreads

The analysis of (36a) is shown in Figure 4.24. The adjunct is a full NP. The schemarequires the adjunct daughter to be fully complete. If it did not have this re-quirement, a noun without determiner like Stunde ‘hour’ in (36b) could enter theschema as adjunct daughter and ungrammatical sentences like (36b) would belicensed.

V[COMPS ⟨⟩]

1 NP

AickeAcike

V[COMPS ⟨ 1 ⟩]

NP[MOD 2 V]

eine Stundeone hour

2 V[COMPS ⟨ 1 ⟩]

liestreads

Figure 4.24: Analysis of an adverbial NP in dass Aicke eine Stunde liest ‘that Aickeis reading for an hour’



4.8 Linking between syntax and semantics

HPSG assumes that all arguments of a head are contained in a list that is calledARGUMENT STRUCTURE (ARG-ST, Davis, Koenig & Wechsler 2021).7 This list con-tains descriptions of the syntactic and semantic properties of the selected argu-ments. For instance the ARG-ST list of English give and its German, Danish andDutch and Icelandic variants is given in (37):

(37) ⟨ NP, NP, NP ⟩

The case systems of the involved languages vary a bit as will be explained inChapter 7, but nevertheless the orders of the NPs in the ARG-ST list are the sameacross these languages. They correspond to nom, dat, acc in German (38a) andsubject, primary object, secondary object in English (38b):

(38) a. dassthat

dasthe

Kindchild

demthe


diethe

Nussnut

gibtgives

‘that the child gives the squirrel the nut’b. that the child gives the squirrel the nut

In addition to the syntactic features we have seen so far semantic features areused to describe the semantic contribution of linguistic objects. (39) shows someaspects of the description of the English verb gives:

(39) lexical item for gives:

ARG-ST⟨NP 1 , NP 2 , NP 3

⟩CONT

giveAGENS 1

GOAL 2

TRANS-OBJ 3

The lowered boxes refer to the referential indices of the NPs. One can imaginethese indices as variables that refer to the object in the real world that the NPis referring to. These indices are identified to semantic roles of the verb give.Finding reasonable role names is not trivial and some authors just use ARG1, ARG2and ARG3 to avoid the problems (see Dowty 1991 for discussion).

7See Pollard & Sag 1994: 28–29; Wechsler 1995; Davis 2001; Müller 2007a: Section 5.6 for argu-ment linking in HPSG. Davis, Koenig & Wechsler (2021) is a handbook article on linking inHPSG.


4.9 Alternatives

The representations for the other languages mentioned above is entirely par-allel. Therefore it is possible to capture crosslinguistic generalizations. Neverthe-less there are differences between the Germanic OV and VO languages. As wasexplained above the VO languages map their subject to SPR and all other argu-ments to COMPS, while the finite verbs of OV languages have all arguments onCOMPS.

4.9 Alternatives

4.9.1 CP/TP/VP models

Grewendorf (1988; 1995), Lohnstein (2014) and many others assume that Germanhas a structure that is parallel to the one that is assumed for English. As forEnglish the verb is assumed to form a phrase with its objects and this VP func-tions as the argument of a Tense head to form a maximal projection togetherwith the subject of the verb, which is realized in the specifier position of the TP.Figure 4.25 shows the analysis of (40) with the respective layers.

(40) dassthat

jedereverbody

diesesthis

Buchbook

kenntknows

‘that everybody knows this book’

The problem with such proposals is that the subject does not depend on theverb but on T. Therefore there is noway of serializing the accusative object beforethe subject unless one assumes that the object is moved to a higher position inthe tree, e.g., adjoined to TP as in Figure 4.26.

While researchers like Frey (1993: 185) argued that quantifier scopings are ac-tually evidence for movement-based approaches, the argument backfires. Let usconsider Frey’s examples. Frey argues that sentences without movement haveonly one reading and sentences like (41b) in which – according to the movement-based theory – movement is involved have two readings: one corresponding tothe visible order and one to the order before movement, the so-called underlyingorder.

(41) a. Esit

istis

nichtnot

derthe

Fall,case

daßthat

erhe

mindestensat.least

einemone

Verlegerpublisher

fastalmost

jedesevery

Gedichtpoem

anbot.offered

‘It is not the case that he offered at least one publisher almost everypoem.’



CP

C′

C

dassthat

TP

NP

jedereverybody

T′

VP

V′

NP

dieses Buchthis book

V

_𝑗

T

kenn-𝑗 -tknow- -s

Figure 4.25: Sentence in the CP/TP/VP model

b. Esit

istis

nichtnot

derthe

Fall,case

daßthat

erhe

fastalmost

jedesevery

Gedicht𝑖poem

mindestensat.least

einemone

Verlegerpublisher

_𝑖 anbot.offered

‘It is not the case that he offered almost every poem to at least onepublisher.’

However, Kiss (2001: 146) and Fanselow (2001: Section 2.6) pointed out that suchapproaches have problems with multiple moved constituents. For instance in anexample such as (42), it should be possible to interpret mindestens einem Verleger‘at least one publisher’ at the position of _𝑖 , which would lead to a reading wherefast jedes Gedicht ‘almost every poem’ has scope over mindestens einem Verleger‘at least one publisher’. However, this reading does not exist.

(42) IchI

glaube,believe

dassthat

mindestensat.least

einemone

Verleger𝑖publisher

fastalmost

jedesevery

Gedicht𝑗poem

nuronly


4.9 Alternatives

CP

C′

C

dassthat

TP

NP𝑖

dieses Buchthis book

TP

NP

jedereverybody

T′

VP

V′

NP

_𝑖

V

_𝑗

T

kenn-𝑗 -tknow- -s

Figure 4.26: Scrambling has to be movement in the CP/TP/VP model

dieserthis

Dichterpoet

_𝑖 _𝑗 angebotenoffered

hat.has

‘I think that only this poet offered almost every poem to at least onepublisher.’

This means that one needs some way to determine the deviation with respectto an unmarked order, but movement is not the solution. See Müller (2020: Sec-tion 3.5) for further discussion and Kiss (2001) for an approach to scope withinthe framework assumed here.



Exercises

1. Provide the valence lists for the following words:

(43) a. laughb. eatc. to doused. bezichtigen

accuse(German)

e. hef. theg. Ankunft

arrival(German)

If you are uncertain as far as case assignment is concerned, you mayuse the Wiktionarya.

2. Draw trees for the NPs that were also used in exercise 1 on page 60 inChapter 3.

(44) a. eineone

Stundehour

vorbefore

derthe

Ankunftarrival

desof.the

Zugestrain


shortlynachafter

derthe

Ankunftarrival

inin

ParisParis


thiseina

Liedsong

singendesinging

Kindchild

ausfrom

demthe

AllgäuAllgäu


3. Draw trees for the following examples. NPs can be abbreviated.


4.9 Alternatives

(45) a. weilbecause

AickeAicke

demthe.DAT

Kindchild

eina.ACC

Buchbook

schenktgives.as.a.present

(German)

‘because Aicke gives the child a book as a present’b. because Kim gave a book to himc. Sandy saw this yesterday.d. at

thatBjarneBjarne

læsteread

bogenbook.DEF

‘that Bjarne read the book’

ahttps://de.wiktionary.org/, 2018-07-02.


https://de.wiktionary.org/

5 The verbal complex

SOV languages like Dutch and German form verbal complexes. There are severalindicators for this that were worked out in detail by Gunar Bech (1955). One wayto analyze such verbal complexes is to assume that the verbs in a sentence forma unit that basically behaves like a simplex verb. This explains for instance whythe arguments of the three verbs in Haider’s example (1986b: 110; 1991: 128) in (1)can be scrambled:

(1) weilbecause

esit

ihrher

jemandsomebody

zuto

lesenread

versprochenpromised

hathas

‘because somebody promised her to read it’

es depends on zu lesen ‘to read’, ihr ‘her’ depends on versprochen ‘promised’ andjemand is the subject and agrees with the finite verb hat ‘has’ (usually it is alsotreated as a dependent of the auxiliary hat).

It should be said that there is extreme variation in the German dialects as far asthe serialization of elements in the verbal complex is concerned. The governingverb is realized to the right of the embedded verb in Standard German: V3 V2 V1as in (1), but there are examples like (3) taken from Müller (1999: 376).1

(2) a. IchI

hättehad

stapelweiseby.the.pile

Aktenfiles

könnencan

haben.have

(German, Berlin dialect)

‘I could have had files by the pile.’b. weil

becauseichI

mirme

dasthat

nichnot

hab’have

lassenlet

gefallenplease

‘because I did not put up with it’c. wenn

ifsethey

mirme

hierhere

würdenwould

rausschmeißen,out.throw

…

‘if they would kick me out here’

The orders in (2) correspond to the order that is most natural in Dutch. (3) showssome Dutch examples:

1Interview partner in: Insekten und andere Nachbarn – ein Haus in Berlin, ARD 15.11.1995.


(3) a. datthat

KimKim

hetthe

boekbook

wilwants

lezenread

’that Kim wants to read the book’b. dat

thatKimKim

SandySandy

hetthe

boekbook

laatlets

lezenread

’that Kim lets Sandy read the book’c. dat

thatKimKim

SandySandy

hetthe

boekbook

wilwants

latenlet

lezenread

’that Kim wants to let Sandy read the book’

SVO languages like Danish and English do not allow the arguments of embed-ded verbs to be scrambled with arguments of higher verbs. All arguments stayin their VP (modulo extraction, of course).

The trick that is used to analyze the verbal complexes is called argument attrac-tion or argument composition and was developed by Geach (1970) in the frame-work of Categorial Grammar and adapted for HPSG by Hinrichs & Nakazawa(1994). The analysis of lesen wird ‘read will’ as it occurs in (4) is shown in Fig-ure 5.1.

(4) dassthat

keinernobody

dasthe

Buchbook

lesenread

wirdwill

‘that nobody will read the book’

V[⟨ NP[nom], NP[acc] ⟩ ]

V[⟨ NP[nom], NP[acc]⟩ ]

lesenread

V[⟨ NP[nom], NP[acc], V ⟩]

wirdwill

Figure 5.1: Analysis of the verbal complex formation of lesen wird ‘read will’ us-ing argument composition (preliminary version)

wird ‘will’ selects an infinitive without zu and in addition its arguments. Thisinfinitive (lesen ‘read’) is combined with the verb and hence is not contained inthe valence list of the mother node.

Returning to our meal-shopping analogy from p. 65, the verbal complex forma-tion can be envisaged by imagining a young and helpful auxiliary verb helping


out a person of the high risk group in the middle of a pandemic. Since high riskpersons are not supposed to do shopping, the helpful person takes over theirshopping list and does the shopping for them. In the case of auxiliary verbs theauxiliary verb just selects the main verb and apart from this does not requireany further arguments apart from the ones taken over from the embedded verb.This means the auxiliary verb just does the shopping for the main verb. A veryaltruistic verb it is. Later we will have a look at verbs like try and let that dorequire their own arguments in addition to the ones of the embedded verb. Thiswill be parallel to a shopping event where the helping person buys its own goodsin addition to buying goods for somebody else.

The combination of lesen and wird behaves like a simplex verb in that it canbe combined with its arguments in any order. Figure 5.2 shows the analysis of(5a) and Figure 5.3 on the following page shows the analysis of (5b).

(5) a. [dass]that

keinernobody

dasthe

Buchbook

lesenread

wirdwill

‘that nobody will read the book’b. [dass]

thatdasthe

Buchbook

keinernobody

lesenread

wirdwill


V[⟨ ⟩]

NP[nom]

keinernobody

V[⟨ NP[nom] ⟩]

NP[acc]

das Buchthe book

V[⟨ NP[nom], NP[acc]⟩]


lesenread


wirdwill

Figure 5.2: Formation of a verbal complex and realization of arguments in normalorder (preliminary version)

I follow Kiss (1995: Section 3.1.1) and represent the subject of non-finite verbsas the value of a special feature SUBJ. SUBJ differs from SPR and COMPS in that it



V[⟨ ⟩]

NP[acc]

das Buchthe book

V[⟨ NP[acc] ⟩]

NP[nom]

keinernobody

V[⟨ NP[nom], NP[acc] ⟩]


lesenread


wirdwill

Figure 5.3: Formation of a verbal complex and scrambling of arguments (prelim-inary version)

is not a valence feature. The reason for this special treatment is that the subjectcannot be realized as a part of a non-finite verb phrase:

(6) a. [Dasthe

Buchbook

lesen]read

wirdwill

AickeAicke

morgen.tomorrow

‘Aicke will read the book tomorrow.’b. * [Aicke

Aickelesen]read

wirdwill

dasthe

Buchbook

morgen.tomorrow

c. ?* [AickeAicke

dasthe

Buchbook

lesen]read

wirdwill

morgen.tomorrow

The lexical item for the non-finite form of lesen ‘to read’ is given in (7):

(7) lesen ‘to read’ non-finite form:[SUBJ ⟨ NP[nom] ⟩COMPS ⟨ NP[acc] ⟩

]The following Attribute Value Matrix (AVM) is a representation of the auxil-

iary verb werden ‘will’:

(8) werden ‘will’ non-finite form:[SUBJ 1

COMPS 2 ⊕ ⟨ V[VFORM bse, LEX+, SUBJ 1 , COMPS 2 ] ⟩

]


werden selects a verb that has the bse form, that is an infinitive without zu ‘to’.The embedded element has to be lexical (LEX+), that is, a single word or a verbalcomplex. All phrases that are licensed by the Head-Complement Schema andthe Specifier-Head Schema are assumed to be LEX−. The boxes with numbersare basically variables. Their values depend on the values of the embedded verbs.Therefore this lexical item can be used with a verb like lesen ‘to read’, whichtakes a nominative and an accusative case but also with a verb like helfen ‘tohelp’, which takes a nominative and a dative object.

Before I turn to the details of the analysis, I have to provide the lexical itemsfor the finite form of auxiliaries. Since the subject of finite verbs can of coursebe realized it has to be represented in one of the valence lists. As was discussedin Section 4.3, German subjects are represented in the COMPS list of finite verbs.Hence the lexical item for wird ‘will’ has the following form:

(9) wird ‘will’ finite form:[SUBJ ⟨⟩COMPS 1 ⊕ 2 ⊕ ⟨ V[VFORM bse, LEX+, SUBJ 1 , COMPS 2 ] ⟩

]This basically says that the valence of wird consists of an embedded verb andwhatever the SUBJ list of this verb is plus whatever the COMPS list of this verb is.This is exemplified for lesen wird in Figure 5.4. The auxiliary selects an infinitive

V[VFORM fin,COMPS 1 ⊕ 2 ]

3 V[VFORM bse,SUBJ 1 ⟨ NP[nom] ⟩,COMPS 2 ⟨ NP[acc] ⟩ ]

lesenread

V[VFORM fin,COMPS 1 ⊕ 2 ⊕ ⟨ 3 ⟩ ]

wirdwill

Figure 5.4: Detailed analysis of a verbal complex

without zu ‘to’ ( 3 ). This is ensured by the value bse for the VFORM feature ofthe selected verb: bse stands for infinitive without to/zu/…, inf stands for aninfinitive form with marker, ppp stands for participle and fin for a finite verb.The subject of the selected infinitive ( 1 ) and the complements ( 2 ) are takenover. The result is that lesen wird has the same arguments as liest ‘reads’.



To make all of this even more fun, we can make it more complex and look atverbal complexes with three verbs. Figure 5.5 shows the analysis of the verbalcomplex lesen können wird ‘read can will’ in sentences like (10):

(10) [dass]that

sieshe

dasthe

Buchbook

lesenread

könnencan

wirdwill


4 V[VFORM bse,SUBJ 1 ,COMPS 2 ]


lesenread

V[VFORM bse,SUBJ 1 ,COMPS 2 ⊕ ⟨ 3 ⟩ ]

könnencan


wirdwill

Figure 5.5: Analysis of a German verbal complex with three verbs in cannonicalorder

One interesting aspect of the analysis is that it can explain a phenomenon thatis called Auxiliary Flip or Oberfeldumstellung. German optionally allows verbsthat govern a modal to be placed to the left of the verbal complex rather than tothe right of the modal. So instead of (10) one can also use the order in (11):

(11) [dass]that

sieshe

dasthe

Buchbook

wirdwill

lesenread

könnencan

After having discussed the analysis of verbal complexes as they are knownfrom the OV languages like German, Dutch, and Afrikaans, I want to brieflycomment on the SVO languages like Danish and English and so on. Usually ahead requires its argument to be saturated, that is, the COMPS value has to bethe empty list for NP, PPs, APs, CPs and sentential and VP arguments. Verbalcomplexes are different: words are combined directly. The VO languages differfrom the OV languages in not allowing this. In VO languages the verb forms




wirdwill

4 V[VFORM bse,SUBJ 1 ,COMPS 2 ]


lesenread

V[VFORM bse,SUBJ 1 ,COMPS 2 ⊕ ⟨ 3 ⟩ ]

könnencan

Figure 5.6: Analysis of a German verbal complex with three verbs with AuxiliaryFlip

a phrase with its complements and this verb phrase may be embedded underanother verb. (12a) shows an example with auxiliary verbs, (12b) is an examplewith a full verb that takes an infinitive verb phrase with to and an object inaddition.

(12) a. Kim [will [have [read the book]]].b. Somebody [promised her [to read it]].

Languages like Danish and English only have the Head-Complement Schemaand the Specifier-Head Schema, while languages like Dutch and German havean additional schema that can combine unsaturated words. The schema for pred-icate complex formation is sketched in Figure 5.7. This schema is very similar

[COMPS 1 ]

2 [COMPS 1 ⊕ ⟨ 2 ⟩]

Figure 5.7: Sketch of the Predicate Complex Schema

to the Head-Complement Schema that was given on page 76. The difference isthat this schema does not license LEX − elements as the Specifier-Head and Head-



Compelement Schema do. Therefore, the combination of two verbs is compatiblewith LEX + requirements by governing verbs and an embedding in even morecomplex verbal complexes is possible. Figure 5.5 is an example: the combinationof lesen ‘read’ and können ‘can’ is compatible with the LEX + requirement of wird‘will’. The Predicate Complex Schema also differs from the Head-ComplementSchema in German in that it combines the head with the last element of the va-lence list. This is the embedded verb and it ha to be combined with the headbefore any other argument since one would not know what the other argumentsare because they are taken over from the embedded verb.

Before turning to the next phenomenon, I want to briefly discuss the alter-native to the verb complex analysis presented here. One alternative suggestionwas to analyze auxiliaries in German as VP embedding verbs (Wurmbrand 2003a).Our standard example would then have the analysis in (13):

(13) dassthat

keinernobody

[[dasthe

Buchbook

lesen]read

wird]will

The question that such analyses have to answer is how scrambling of argumentsof the involved verbs can be accounted for. The answer is often that it is assumedthat the object of the embedded verb is extracted from the VP and moved to theleft periphery of the clause. This is shown in (14):

(14) dassthat

[dasthe

Buch]𝑖book

keinernobody

[[ _𝑖 lesen]read

wird]will


However, analyses that treat scrambling as movement are problematic since theypredict additional readings of sentences that have quantifiers in their NPs (Kiss2001: 146; Fanselow 2001: Section 2.6).

Before I turn to the analysis of the verb position, I want to show how sentenceswith several verbs in SVO languages can be analyzed. Figure 5.8 on the facingpage shows the analysis of the English version of sentence (5a). The verb readsselects a subject and an object. The verb forms a VP with the NP the book. ThisVP is still lacking a subject. The auxiliary will selects a VP and a subject that isidentical with the subject of read. The combination of will and the VP is licensedby the Head-Complement Schema that was sketched in Figure 4.11 on page 76.

The equivalent of lesen können wird ‘read can will’ cannot be given here, sinceEnglish modal verbs do not have non-finite forms, but one can construct exam-ples with modals as the highest verb:

(15) She [must [have [seen it]]].


S

NP[nom]

nobody

VP

V

will

VP

V

read

NP[acc]

the book

Figure 5.8: Embedding of a VP in SVO languages

This sentence has a structure that is similar to the one in Figure 5.8: must andhave both embedd VPs.

Finally, Figure 5.9 on the following page shows the translation of (1):

(16) Somebody has promised her to read it.

promise is a verb that takes a subject, an object, and a VP complement. Like inthe analysis of (20) on page 76 – which is repeated here as (17) for convenience– the verb promised is combined with its NP complement first and then with itsVP argument.

(17) Nobody gave the child a book.

The VP argument of promised in Figure 5.9 consists of to and another VP with aninfinitive in base form. to is analyzed as an auxiliary verb (Gazdar et al. 1982: 600;Sag et al. 2020: 147). It is important to note that the object him cannot appear inany other position (appart from extraction to the left periphery). For instance, itcannot appear in the position of the book and the same holds for the book: Thisphrase cannot appear in any other place than in the object position.

Exercises

1. Sketch the analysis of the verbal complexes in the following examples:



S

NP[nom]

somebody

VP

V

has

VP

V

V

promised

NP[acc]

her

VP

V

to

VP

V

read

NP[acc]

the book

Figure 5.9: Embedding of a VP with verbs that take an additional object

(18) a. dassthat

sieshe

darüberthere.about

lachenlaugh

mussmust

(German)

‘that she has to laugh about it’b. dass

thatsieshe

darüberthere.about

hathas

lachenlaugh

müssenmust

‘that she had to laugh about it’c. dass

thatsieshe

darüberthere.about

wirdwill

habenhave

lachenlaugh

müssenmust

You may omit the SPR values, since they are the empty list for all Ger-man verbs anyway.

2. Search for two sentences with a verbal complex in a newspaper or incorpora (for example the COSMAS corpusa) and analyze the verbal com-plexes.

3. Search for verbal complexes with more than four verbs in a corpus and


document your search.ahttps://cosmas2.ids-mannheim.de, 2020-05-11.

Further reading

The analysis of verbal complexes in HPSG was first developed by Hinrichs& Nakazawa (1989a,b). Hinrichs & Nakazawa (1994) is the first peer re-viewed publication on this topic by Hinrichs & Nakazawa. Kiss (1995) is amonograph dealing with verbal complexes. Meurers (2000) also deals withverbal complexes including difficult cases like the so-called Zwischenstel-lung, which is no treated here. Müller (2002) treats not just verbal complexbut also other types of complex predicates like adjective verb complexes,resultative constructions and particle verbs.

Godard & Samvelian (2021) provide an overview about analyses of com-plex predicates in HPSG.


https://cosmas2.ids-mannheim.de

6 Verb position: Verb first and verbsecond

This chapter deals with the analysis of the verb position in V2 langauges. I willconcentrate on Danish and German, which may serve as prototypical examples:Danish is an SVO language, while German is SOV. I will first discuss argumentsfor the classification of German as an SOV language and provide the necessarydata on Danish and then explain the respective analyses.

6.1 The phenomenon

Section 2.1 contains a discussion of the basic order of subject, object and verbin the languages of the world and in the Germanic languages in particular. Idiscussed the classification provided by the World Atlas of Language Structures,which suggested that German is a language with two dominant constituent or-ders: SOV and SVO. Claiming that SVO is a basic order on the basis of purecounting is somehow strange given the fact that most German sentences do nothave the subject in first position anyway. The following text may serve as anexample:

(1) Für selbstfahrendeAutos soll es in Deutschland nachAngaben von Bundes-verkehrsminister Alexander Dobrindt (CSU) bald eine Teststrecke geben.Auf der Autobahn A9 in Bayern sei ein Pilotprojekt „Digitales TestfeldAu-tobahn“ geplant, wie aus einem Papier des Bundesverkehrsministeriumshervorgeht. Mit den ersten Maßnahmen für diese Teststrecke solle schonin diesem Jahr begonnenwerden. Mit dem Projekt soll die Effizienz vonAu-tobahnen generell gesteigert werden. „Die Teststrecke soll so digitalisiertund technisch ausgerüstet werden, dass es dort zusätzliche Angebote derKommunikation zwischen Straße und Fahrzeug wie auch von Fahrzeugzu Fahrzeug geben wird“, sagte Dobrindt zur Frankfurter AllgemeinenZeitung. Auf der A9 sollten sowohl Autos mit Assistenzsystemen als auchspäter vollautomatisierte Fahrzeuge fahren können. Dort soll die Kom-munikation nicht nur zwischen Testfahrzeugen, sondern auch zwischen

6 Verb position: Verb first and verb second

Sensoren an der Straße und den Autos möglich sein, etwa zur Übermit-tlung von Daten zur Verkehrslage oder zum Wetter. Das Vorhaben solleim Verkehrsministerium von einem runden Tisch mit Forschern und In-dustrievertretern begleitet werden, sagte Dobrindt. Dieser solle sich unteranderem auch mit den komplizierten Haftungsfragen beschäftigen. Also:Wer zahlt eigentlich, wenn ein automatisiertes Auto einen Unfall baut?[Mithilfe der Teststrecke] solle die deutsche Automobilindustrie auch beimdigitalen Auto „Weltspitze sein können“, sagte der CSU-Minister. Die deut-schenHersteller sollten die Entwicklung nicht Konzernenwie etwaGoogleüberlassen. Derzeit ist Deutschland noch an das „Wiener Übereinkommenfür den Straßenverkehr“ gebunden, das Autofahren ohne Fahrer nicht zulässt. Nur unter besonderen Auflagen sind Tests möglich. Die Grünen hal-ten die Pläne für unnütz. Grünen-Verkehrsexpertin Valerie Wilms sagteder Saarbrücker Zeitung: „Der Minister hat wichtigere Dinge zu erledigen,als sichmit selbstfahrenden Autos zu beschäftigen.“ Die Technologie sei imVerkehrsbereich nicht vordringlich, auch stehe sie noch ganz am Anfang.Aus dem grün-rot regierten Baden-Württemberg – mit dem Konzernsitzvon Daimler – kamen hingegen andere Töne. Was in Bayern funktioniere,müsse auch in Baden-Württemberg möglich sein, sagte Wirtschaftsminis-ter Nils Schmid (SPD). Von den topografischen Gegebenheiten biete sichdie Autobahn A81 an. (taz: 27.01.2015)

The subjects are marked in red and the non-subjects in green. I also countedsubjects/non-subjects within embedded clauses. The ratio is 10 subjects com-pared to 15 non-subjects. So, the question is: What does this number tell us?Of course we could now further differentiate the grammatical functions of thefronted material. We would find that we have 3 object clauses fronted, the rest ofthe fronted constituents is adverbials. We could conclude that SVO is more com-mon than OVS, but saying that SVO is basic would not be appropriate. RatherAdv V S O should be regarded as a basic pattern if we assume this little text asour empirical basis. Of course, assuming the two texts as the basis of scientificclaims is not sufficient. Hinrichs & Kübler (2005: Section 4) examined the Vorfeldconstituents in the TüBa-D/S and Z corpora. The S corpus contains spoken Ger-man from the machine translation project Verbmobil and the Z corpus sentencesfrom the German newspaper taz. Both corpora are annotated for grammaticalfunction. The TüBa-D/S consisted of a total of 38,342 trees and the TüBa-D/Ztreebank had 22,087 trees when the paper was written in 2005. The two corporahad subjects in the Vorfeld in 50.3 % and 52.1 % of the cases, respectively. So, as-suming SVO as the basic order would not be helpful, since in about 50 % of the


6.1 The phenomenon

clauses and may be even more, one would have to deal with an order in whichthe subject is not in initial position. On top of this, there would be the prob-lem of subordinated sentences, which clearly do have a SOV order. Thereforesyntacticians of various different frameworks (see Müller 2020 for approachesin GB, Minimalism, LFG, Categorial Grammar, and HPSG) assume that SOV isthe base order of German. The finite verb is fronted to mark the sentence typeand one constituent is put in front of this verb. This fronted constituent can bethe subject, an object or any other constituent of the sentence. It may be even adependent of a deeply embedded element in the clause. So, the position infrontof V in the V2 languages has nothing to do with the SVO/SOV dichotomy andbasically disturbs the picture and makes the counting-approach pursued in theWALS (see Section 2.1) non-applicable.

In the following I will provide facts that are seen as evidence for SOV as thebasic order of German (and Dutch). Before I provide an analysis in Section 6.2,I discuss the verb position in the germanic SVO languages with Danish as anexample in Section 6.1.2.

6.1.1 German as an SOV language

6.1.1.1 The order of particle and verb and idioms

Verb particles form a close unit with the verb. The unit is observable in verb finalsentences only, which supports an SOV analysis (Bierwisch 1963: 35).

(2) a. weilbecause

erhe

morgentomorrow

anfängtat.catches

‘because he starts tomorrow’b. Er

hefängtcatches

morgentomorrow

an.at

‘He starts tomorrow.’

The particle verb in (2) is non-transparent: its meaning is not related to the verbfangen ‘to catch’. Such particle verbs are sometimes called mini idioms. In factthe argument above can also bemadewith real idioms: Many idioms do not allowrearrangement of the idiom parts:

(3) a. dassthat

niemandnobody

demthe

Mannman

denthe

GarausGARAUS

machtmakes

‘that nobody kills the man’



b. ?* dassthat

demthe

Mannman

denthe

GarausGARAUS

niemandnobody

machtmakes

c. Niemandnobody

machtmakes

ihmhim

denthe

Garaus.GARAUS

‘Nobody kills him.’

This is an instance of Behaghel’s law (1932) that things that belong together se-mantically tend to be realized together. The exception is the finite verb. Thefinite verb can be realized in initial or final position despite the fact that thisinterrupts the continuity of the idiomatic material. Since the continuity can beobserved in SOV order only, this order is considered basic.

6.1.1.2 Verbs formed by back-formation

Verbs that are derived from nouns by backformation often cannot be separatedand verb second sentences therefore are excluded (seeHaider 1993: 62, who refersto unpublished work by Höhle 1991):

(4) a. weilbecause

siethey

dasthe

Stückplay

heutetoday

uraufführenplay.for.the.first.time

‘because they premiered the play today’b. * Sie

theyuraufführenplay.for.the.first.time

heutetoday

dasthe

Stück.play

c. * Siethey

führenguide

heutetoday

dasthe

Stückplay

urauf.PREFIX.PART

Hence these verbs can only be used in the order that is assumed to be the baseorder.

6.1.1.3 Constructions that only allow SOV order

Similarly, it is sometimes impossible to realize the verb in initial position whenelements like mehr als ‘more than’ are present in the clause (Haider 1997b; Mei-nunger 2001):

(5) a. dassthat

HansHans

seinenhis

Profitprofit

letzteslast

Jahryear

mehrmore

alsthan

verdreifachtetripled

‘that Hans increased his profit last year by a factor greater thanthree’


6.1 The phenomenon

b. HansHans

hathas

seinenhis

Profitprofit

letzteslast

Jahryear

mehrmore

alsthan

verdreifacht.tripled

‘Hans increased his profit last year by a factor greater than three.’c. * Hans

Hansverdreifachtetripled

seinenhis

Profitprofit

letzteslast

Jahryear

mehrmore

als.than

So, it is possible to realize the adjunct together with the verb in final position, butthere are constraints regarding the placement of the finite verb in initial position.

6.1.1.4 Order in subordinate and non-finite clauses

Verbs in non-finite clauses and in subordinate finite clauses starting with a con-junction always appear finally, that is, in the rechte Satzklammer. For example,zu geben ‘to give’ and gibt ‘gives’ appear in the rechte Satzklammer in (6a) and(6b):

(6) a. Derthe

Clownclown

versucht,tries

Kurt-MartinKurt-Martin

diethe

Waregoods

zuto

geben.give

‘The clown tries to give Kurt-Martin the goods.’b. dass

thatderthe

Clownclown

Kurt-MartinKurt-Martin

diethe

Waregoods

gibtgives

‘that the clown gives Kurt-Martin the goods’

6.1.1.5 Scope of adverbials

The scope of adverbials in sentences like (7) depends on their order (Netter 1992:Section 2.3): The left-most adverb scopes over the following adverb and over theverb in final position. This was explained by assuming the following structure:

(7) a. weilbecause

erhe

[absichtlichdeliberately

[nichtnot

lacht]]laughs

‘because he deliberately does not laugh’b. weil

becauseerhe

[nichtnot


lacht]]laughs

‘because he does not laugh deliberately’

An interesting fact is that the scope relations do not change when the verb posi-tion is changed. If one assumes that the sentences have an underlying structurelike in (7), this fact is explained automatically:



(8) a. Lacht𝑖laughs

erhe


[nichtnot

_𝑖]]?

‘Does he deliberately not laugh?’b. Lacht𝑖

laughserhe

[nichtnot


_𝑖]]?

‘Doesn’t he laugh deliberately?’

It has to be mentioned here, that there seem to be exceptions to the claim thatmodifiers scope from left to right. Kasper (1994: 47) discusses the examples in (9),which go back to Bartsch & Vennemann (1972: 137).

(9) a. PeterPeter

liestreads

wegenbecause.of

derthe

Nachhilfestundentutoring

gut.well

‘Peter reads well because of the tutoring.’b. Peter

Peterliestreads

gutwell

wegenbecause.of

derthe

Nachhilfestunden.tutoring

(9a) corresponds to the expected order in which the adverbial PPwegen der Nach-hilfestunden outscopes the adverb gut, but the alternative order in (9b) is possibleas well and the sentence has the same reading as the one in (9a).

However, Koster (1975: Section 6) and Reis (1980: 67) showed that these exam-ples are not convincing evidence since the rechte Satzklammer is not filled andtherefore the orders in (9) are not necessarily variants of Mittelfeld orders butmay be due to extraposition of one constituent. As Koster and Reis showed, theexamples become ungrammatical when the right sentence bracket is filled:

(10) a. * HansHans

hathas

gutwell

wegenbecause.of

derthe

Nachhilfestundentutoring

gelesen.read

b. HansHans

hathas

gutwell

gelesenread

wegenbecause.of

derthe

Nachhilfestunden.tutoring

‘Peter read well because of the tutoring.’

The conclusion is that (9b) is best treated as a variant of (9a) in which the PP isextraposed.

While examples like (9) show that the matter is not trivial, the following ex-ample from Crysmann (2004: 383) shows that there are examples with a filledrechte Satzklammer that allow for scopings in which an adjunct scopes over an-other adjunct that precedes it. For instance, in (11) niemals ‘never’ scopes overwegen schlechten Wetters ‘because of the bad weather’:


6.1 The phenomenon

(11) Dathere

mußmust

esit

schonPART

erheblichesevere

Problemeproblems

mitwith

derthe

Ausrüstungequipment

gegebengiven

haben,have

dasince

[wegenbecause.of

schlechtenbad

Wetters]weather

eina

ReinholdReinhold

MessnerMessner

[niemals]never

aufgäbe.give.up.would

‘There must have been severe problems with the equipment, sincesomeone like Reinhold Messner would never give up just because of thebad weather.’

However, this does not change the fact that the sentences in (7) and (8) havethe same meaning independent of the position of the verb. The general meaningcomposition may be done in the way that Crysmann suggested.

Another word of caution is in order here: there are SVO languages like Frenchthat also have a left to right scoping of adjuncts (Bonami et al. 2004: 156–161). So,the argumentation above should not be seen as the only fact supporting the SOVstatus of German. In any case the analyses of German that were worked out invarious frameworks can explain the facts nicely.

6.1.1.6 Position of non-finite verbs in VO and OV languages

Before I turn to the verb position in Danish in the next subsection, I want torepeat Ørsnes’ examples containing several non-finite verbs: the example in (12a)shows a German subordinate clause with a verbal complex consisting of threeverbs. The level of embedding is indicated by subscript numbers. As can be seen,the verbs are added at the end of the clause. In the corresponding Danish examplefrom Ørsnes (2009: 146), it is exactly the other way around: the embedding verbpreceeds the embedded verb.

(12) a. dassthat

erhe

ihnhim

gesehen3seen

haben2have

muss1must

(German)

‘that he must have seen him’b. at

thathanhe

må1must

have2have

set3seen

hamhim

(Danish)

The examples in (13) shows variants with different complexity. If we exchangethe simplex verb sah ‘saw’ in (13a) by the perfect form, the auxiliary is placedafter the participle as in (13b).



(13) a. dassthat

erhe

ihnhim

sahsaw

(German)

‘that he saw him’b. dass

thaterhe

ihnhim

gesehenseen

hathas

‘that he has seen him’

If a modal is added to (13b), the modal goes to the right of the embedded verbs.This order is distorted by the placement of the finite verb in initial position, butthis placement is independent of the order of the non-finite verbs. As the ex-amples in (14) show, the finite verb is realized to the left of the subject both inGerman (SOV) and in Danish (SVO).

(14) a. Mussmust

erhe

ihnhim

gesehenseen

haben?have

(German)

‘Must he have seen him?’b. Må

musthanhe

havehave

setseen

ham?him

(Danish)

‘Must he have seen him?’

6.1.2 Verb position in the Germanic SVO languages

During the discussion of scope facts I already hinted at an analysis in whicha trace marks the position of the verb in final position and the verb in initialposition is coindexed with this trace. Although the SVO languages are differenta similar analysis has been suggested for languages like Danish. The evidencefor this is that adverbials in SVO languages usually attach to the VP, that is theycombine with a phrase consisting of the verb and its object or objects. (15) givesan example:

(15) atthat

ConnyConny

ikkenot

[VP læserreads

bogen]book.DEF

(Danish)

‘that Conny does not read the book’

The interesting thing now is that the finite verb is placed to the left of thenegation in V2 sentences:

(16) ConnyConny

læserreads

ikkenot

bogen.book.DEF

(Danish)

‘Conny is not reading the book.’


6.1 The phenomenon

This is seen as evidence for verb fronting by many:

(17) ConnyConny

læser𝑖reads

ikkenot

[VP _𝑖 bogen].book.DEF

‘Conny does not read the book.’

With this as a background it should be clear what the analysis of yes/no ques-tions as the one in (18b) is:

(18) a. atthat

ConnyConny

læserreads

bogenbook.DEF

‘that Conny reads the book’b. Læser

readsConnyConny

bogen?book.DEF

‘Does Conny read the book?’

The analysis of the first sentence involves a VP as in (19a) and the second sentenceinvolves a VP with a verbal trace that corresponds to the verb in initial position:

(19) a. atthat

ConnyConny

[VP læserreads

bogen]book.DEF

‘that Conny reads the book’b. Læser𝑖

readsConnyConny

[VP _𝑖 bogen]?book.DEF


It is interesting to note that the German and the Danish question with simplexverbs have exactly the same constituent order. Compare (18b) with (20):

(20) Liestreads

ConnyConny

dasthe

Buch?book

(German)


The internal structure of these sentences is quite different though. The differentnature of the two langauges is of course more obvious when non-finite verbs areinvolved:

(21) a. Har𝑖has

ConnyConny

[ _𝑖 læstread

bogen]?book.DEF

(Danish)

‘Has Conny read the book?’



b. Hat𝑖has

ConnyConny

dasthe

Buchbook

[gelesenread

_𝑖]? (German)

‘Has Conny read the book?’

In (21a) the finite verb is connected to a trace in initial position of the VP and in(21b) it is connected to a verb in final position in a verbal complex.

6.1.3 Verb second

Even languages with rather rigid constituent order sometimes allow to front ele-ments or to position elements at the far right, that is, extrapose them. (22) showsEnglish examples of fronting:

(22) a. I read this book yesterday.b. This book, I read yesterday.c. Yesterday, I read this book.

The object this book and the adjunct yesterday are fronted in (22b) and (22c),respectively.

The Germanic languages (with the exception of English) place one constituentin front of the finite verb. As the German examples in (23) show, the frontedconstituent can be of any grammatical function:

(23) a. IchI

habehave

dasthe

Buchbook

gesternyesterday

gelesen.read

(German)

‘I have read the book yesterday.’b. Das

theBuchbook

habehave

ichI

gesternyesterday

gelesen.read

c. Gesternyesterday

habehave

ichI

dasthe

Buchbook

gelesen.read

d. Gelesenread

habehave

ichI

dasthe

Buchbook

gestern,yesterday

gekauftbought

hattehad

ichI

esit

aberbut

schonyet

vorbefore

einema

Monat.month

‘I read the book yesterday, but I bought it last month already.’e. Das

theBuchbook

gelesenread

habehave

ichI

gestern.yesterday


6.1 The phenomenon

Such frontings are not clause-bound, that is the fronting may cross one or sev-eral clause boundaries and also boundaries of other constituents. (24) shows En-glish examples in which the object of saw is extracted across one and two clauseboundaries:

(24) a. Chris, Sandy saw.b. Chris, we think that Sandy saw.c. Chris, we think Anna claims that Sandy saw.

In German such extractions can be found as well:

(25) a. Wen𝑖who

glaubstbelieves

du,you

daßthat

ichI

_𝑖 gesehenseen

habe.1

have(German)

b. „Wer𝑖 , glaubt er, daß er _𝑖 ist?“ erregte sich ein Politiker vom Nil.2

It is generally said that they are more common in Southern German variaties,but there are other examples that show that nonlocal dependencies are involved.In (26a) the prepositional object um zwei Millionen Mark ‘around two millionDeutscheMarks’ depends on betrügen ‘to cheat’. It does not depend on any of theverbs in thematrix clause. The phrase eine Versicherung zu betrügen ‘an insuranceto betray’ is extraposed that is it is positioned to the right of the verbal braketin the so-called Nachfeld. The position of um zwei Millionen Mark cannot beaccounted for by local reordering. Similarly, gegen ihn ‘against him’ dependson Angriffe ‘attacs’, which is part of the phrase Angriffe zu lancieren ‘attacs tolaunch’. Again an analysis based on local reordering of dependents of a head isimpossible.

(26) a. [Umaround

zweitwo

Millionenmillion

Mark]𝑖Deutsche.Marks

sollshould

erhe

versuchttried

haben,have

[einean

Versicherunginsurance.company

_𝑖 zuto

betrügen].3

deceive‘He apparently tried to cheat an insurance company out of twomillion Deutsche Marks.’

b. [Gegenagainst

ihn]𝑖him

fallefall

esit

denthe

RepublikanernRepublicans

hingegenhowever

schwerer,more.difficult

1Scherpenisse (1986: 84).2Spiegel, 8/1999, S. 18.3taz, 04.05.2001, p. 20.



[ [ Angriffeattacks

_𝑖] zuto

lancieren].4

launch‘It is, however, more difficult for the Republicans to launch attacksagainst him.’

6.2 The analysis

This section deals with verb first sentences in Subection 6.2.1 and then coversverb second sentences in Subsection 6.2.2.

6.2.1 Verb first

The analysis uses a mechanism that passes up information in a tree. The verbaltrace contains the information that a verb is missing locally. This informationabout the missing verb is passed up to the node that dominates the verbal trace.It is represented using // (read double slash). The respective information is head-information and therefore it is passed up the head-path along with other infor-mation as for instance part of speech. Figure 6.1 illustrates. The verbal trace ismissing a V, the V is missing a V and the S as well. The initial verb selects for

S

V⟨S//V

⟩V

liest𝑗reads

S//V

NP

ConnyConny

V′//V

NP

das Buchthe book

V//V

_𝑗

Figure 6.1: Analysis of verb position in German

a sentence that is lacking a V⟨S//V

⟩. The lexical item for the verb in initial

position is licensed by a lexical rule that relates a verb to a verb that selects for a

4taz, 08.02.2008, p. 9.


6.2 The analysis

sentence that is lacking the input verb. Since the selectional requirement of thisverb (S//V) is identified with the sentence lacking a V (Conny das Buch _𝑗 ), theinformation about the original verb liest is identified with the V in S//V. Sincethe double slash information is head information, it percolates down along thehead path to the verbal trace. The information about the initial V is identifiedwith the syntactic and semantic information of the verbal trace in final positionand hence this verbal trace behaves exactly like the verb in inital position thatwas input to the lexical rule.

Various researchers argued that the finite verb in initial position behaves likea complementizer in subordinated clauses (Höhle 1997; Weiß 2005; 2018). This iscaptured by the analysis. Compare Figure 6.1 with Figure 6.2. The complemen-

CP

C ⟨ S ⟩

dassthat

S

NP

ConnyConny

V′

NP

das Buchthe book

V

liestreads

Figure 6.2: Analysis of a verb final clause with complementizer in German

tizer dass ‘that’ selects for a complete sentence, that is, a sentence that does nothave a missing verb, and the initial verb liest ‘reads’ in Figure 6.1 selects for a sen-tence that is missing liest. So apart from the overt or covert verb the structuresare identical. This fact is important when it comes to the analysis of the scopefacts. Since the structure is completely parallel to the one we have in verb finalsentences, the scope facts follow immediately: the trace behaves like the verbin initial position, absichtlich ‘deliberately’ modifies the trace and the resultingsemantics is passed up in the tree (see Figure 6.3 on the next page). The nextstep is the modification by nicht ‘not’. Again the resulting semantics is passedup. lacht ‘laughs’ combines with the clause and takes its semantics over. Sincelacht is the head the semantics is passed on from there.

The analysis of Danish is completely parallel to the one of German. The only



S

V⟨S//V

⟩V

lacht𝑗laughs

S//V

NP

erhe

V′//V

Adv

nichtnot

V′//V

Adv

absichtlichdeliberately

V//V

_𝑗

Figure 6.3: Analysis of sentences with adverbials in German

difference between Figure 6.1 and Figure 6.4 is the position of the verbal tracerelative to the object: the trace follows the object in German and it preceeds it inDanish.

S

V⟨S//V

⟩V

læser𝑗reads

S//V

NP

ConnyConny

VP//V

V//V

_𝑗

NP

bogenbook.DEF

Figure 6.4: Analysis of verb position in Danish

The last thing that is explained in this chapter is the analysis of negation andverb fronting in Danish. Figure 6.5 on the next page shows that the negation at-


6.2 The analysis

taches to the VP as in verb final clauses and the verb is fronted so that it appears tothe left of the negation. The next chapter explains the extraction of constituents

S

V⟨S//V

⟩V

læser𝑗reads

S//V

NP

ConnyConny

VP//V

Adv

ikkenot

VP//V

V//V

_𝑗

NP

bogenbook.DEF

Figure 6.5: The analysis of verb fronting and negation in Danish

and it will then be possible to provide the full structure for sentences like (27a)and it will become clear why the order of negation and verb differs in embeddedand main clauses:

(27) a. ConnyConny

læserreads

ikkenot

bogen.book.DEF

‘Conny does not read the book.’b. at

thatConnyConny

ikkenot

læserreads

bogenbook.DEF

‘that Conny does not read the book’

6.2.2 Verb second

The technique that is used for the analysis of nonlocal dependencies is the samethat was employed for the analysis of the reorderings of verbs: an empty elementtakes the position of the fronted constituent and the information about the miss-ing constituent (the so-called gap) is passed up in the tree until it is finally boundoff by the fronted element, the so-called filler. Figure 6.6 on the following pageillustrates the analysis of (24a).



S

NP

Chris

S/NP

NP

Sandy

VP/NP

V

saw

NP/NP

_

Figure 6.6: The analysis of extraction in English

Figure 6.7 on the next page shows the analysis of example (24b), which re-ally requires a nonlocal dependency. As is shown in the figure, the informationabout the missing object is passed up to the sentence level (S/NP), to the CP level(CP/NP) and up to the next higher S. There it is bound off by the filler Chris. Thebinding off of the missing element is licensed by a special schema, which is calledthe Filler-Head Schema. Figure 6.8 on the facing page provides a sketch of thisschema.

English is the only non-V2 language among the Germanic languages. In whatfollows I show how German (V2+SOV) and Danish (V2+SVO) can be analyzedwith the techniques that were introduced so far. Figure 6.9 on page 124 showsthe analysis of (28):

(28) Dasthe

Buchbook

liestreads

Conny.Conny

‘Conny reads the book.’

The analysis of the German example is more complicated than the English onesince verb movement is involved. The verb is fronted as was explained withreference to Figure 6.1. In addition the object is realized by a trace and then filledby the filler das Buch ‘the book’, which is realized preverbally.

I follow Fanselow (2003) and Frey (2004), who assume that the position ofthe object is initial in the Mittelfeld. Since German allows for both nominative,accusative and accusative, nominative order, the position of the trace for theextracted object could be initial or final as in (29a) and (29b), respectively:


6.2 The analysis

S

NP

Chris

S/NP

NP

we

VP/NP

V

think

CP/NP

C

that

S/NP

NP

Sandy

VP/NP

V

saw

NP/NP

_

Figure 6.7: Extraction crossing the clause boundary

H

1 H/ 1

Figure 6.8: Sketch of the Head-Filler Schema

(29) a. [Dasthe

Buch]𝑖book

liest𝑗reads

_𝑖 ConnyConny

_𝑗 .

b. [Dasthe

Buch]𝑖book

liest𝑗reads

ConnyConny

_𝑖 _𝑗 .

Fanselow and Frey refer to information structural properties that elements in theinitial position have and argue that fronted elements like das Buch have informa-tion structural properties that correspond to the ones that non-fronted elementsin the initial Mittelfeld position have:



S

NP𝑖

das Buchthe book

S/NP

V⟨S//V

⟩V

liest𝑗reads

S//V/NP

NP/NP

_𝑖

V′//V

NP

ConnyConny

V//V

_𝑗

Figure 6.9: Analysis of V2 in German (SOV)

(30) Liestreads

dasthe

Buchbook

Conny?Conny

‘Does Conny read the book.’

They argue that (30) patterns with (29a) rather than with (29b).The complete discussion will not be repeated here, since this would take us

too far away, but the interested reader may consult the discussion section 4.9.1.The Danish example is similar. We first have the analysis of verb-initial posi-

tion that involves the double slash mechanism and on top of that we have thefronting of the object using the slash mechanism. Figure 6.10 on the facing pageillustrates.

The careful reader will ask why we use two different mechanisms to analyzeverb movement and extraction. The answer is that these movement types aredifferent in nature: Verb movement is clause bound while the movement of otherconstituents may cross clause-boundaries. This is captured by the fact that thedouble slash information is passed up together with other head features as forinstance the part of speech information and the slash information is passed upseparately.

Before we deal with passive in the next chapter, we can compare the threesentences in (31):

(31) a. Conny reads a book.


6.2 The analysis

S

NP𝑖

bogenbook.DEF

S/NP

V⟨S//V

⟩V

læser𝑗reads

S//V/NP

NP

ConnyConny

VP//V/NP

V//V

_𝑗

NP/NP

_𝑖

Figure 6.10: Analysis of V2 in Danish (SVO)

b. Conny læser en bog.c. Conny liest ein Buch.

Again the order of the elements is the same in all three languages. However,English is an SVO non-V2 language, Danish is an SVO+V2 language and Germanis an SOV+V2 language. The analyses in bracket notation are given in (32), thetree structures are depicted in Figure 6.11 on the next page:

(32) a. [S Conny [VP reads [NP a book]]].b. [S Conny𝑖 [S/NP læser𝑗 [S/NP _𝑖 [VP _𝑗 [NP en bog]]]].c. [S Conny𝑖 [S/NP liest𝑗 [S/NP _𝑖 [V [NP ein Buch] _𝑗 ]]]].

It may be surprising that these three sentences get such radically different analy-ses although the order of elements are the same. The difference in structures isthe result of the assumption that all declarative main clauses in the GermanicV2 languages follow the same pattern, namely that the finite verb is fronted andthen another constituent is fronted. This particular construction is connected tothe clause type, that is, to the meaning of the utterance (imperative, question,assertion). The sentences in (25) and (26) show that V2 involves a nonlocal de-pendency. Therefore the analysis of (32b) is more complex than (33) and involvesthe fronting of the finite verb to intital position with a successive fronting of thesubject:



S

NP

Conny

VP

V

reads

NP

a book

S

NP𝑖

ConnyConny

S/NP

V⟨S//V

⟩V

læser𝑗reads

S//V/NP

NP/NP

_𝑖

VP//V

V//V

_𝑗

NP

en boga book

S

NP𝑖

ConnyConny

S/NP

V⟨S//V

⟩V

liest𝑗reads

S//V/NP

NP/NP

_𝑖

V′//V

NP

ein Bucha book

V//V

_𝑗

Figure 6.11: Declarative main clauses with subject in initial position in English,Danish and German: despite similar appearance, the syntactic struc-ture is different

(33) [S Conny [VP læser [NP en bog]]].

The reason is that now all declarative main clauses are subsumed under the samestructure, namely (32b). A declarative main clause in all Germanic V2 languagesis the combination of an extracted phrase with a verb initial phrase in whichthe extracted element is missing. Fronting of the finite verb is a way to mark theclause type: if just the finite verb is fronted, a yes/no question (34a) or imperativeresults (34b).5

(34) a. Gibtgives

erhe

ihmhim

dasthe

Buch?book

(German)

‘Does he give him the book?’b. Gib

givemirme

dasthe

Buch!book

If another constituent is fronted, a question with question word (35a), an imper-

5Verb initial clausesmay also be declarative clauses if so-called topic drop (Fries 1988) is involved:

(i) Waswhat

machtdoes

Peter?Peter

Gibtgives

ihmhim

eina

Buch.book

(German)

‘What does Peter do? He gives him a book.’

The subject of gibt ‘gives’ is dropped. The complete sentence would be a V2 sentence: Er gibtihm ein Buch..


6.2 The analysis

ative (35b) or a declarative clause (35c) results.

(35) a. Wemwho

gibtgives

erhe

dasthe

Buch?book

(German)

‘Whom does he give the book to?’b. Jetzt

nowgibgive

ihmme

dasthe

Buch!book

‘Give me the book now!’c. Jetzt

nowgibtgives

erhe

ihmhim

dasthe

Buch.book

‘He gives him the book now.’

The analysis of the semantics of clause types cannot be given here but the inter-ested reader is referred to Müller (2015b; 2021b).


1. How are clause types determined in the Germanic languages?

Exercises

1. Classify the Germanic languages according to their basic constituentorder (SVO, SOV, VSO, …) and V2 assuming that you know that one ofthe following patterns exist in the language:

(36) a. NP[acc] V-Aux NP[nom] V NP[dat]b. NP[acc] V-Aux NP[nom] NP[dat] Vc. NP[acc] NP[nom] V NP[acc]d. NP[acc] NP[nom] V-Aux V NP[acc]e. NP[acc] V-Aux NP[nom] V PP

Every sentence should be paired with ±V2 and one of the six permuta-



tions of S, O, and V.

If you cannot determine the order unambiguously, please say so. Ifyou think that this pattern does not exist in any of the Germanic lan-guages say so. Please keep in mind that English is a so-called residualV2 language, which means that there are some traces of V2 left in thegrammar. Think about question formation in English.

2. Sketch the analysis for the following examples. Use the abbreviationsused in this chapter, that is, do not go into the details regarding SPR andCOMPS values but use S, VP, and V′. Verb movement should be indicatedwith the ‘//’ symbol.

(37) a. Arbejderworks

BjarneBjarne

ihærdigtseriously

påat

bogen?book.DEF

(Danish)

‘Does Bjarne work seriously on the book?’b. Arbeitet

worksBjarneBjarne

ernsthaftseriously

anat

demthe

Buch?book

(German)

‘Does Bjarne work seriously on the book?’c. Wird

willsieshe

darüberthere.upon

nachdenken?think

(German)

‘Will she think about this?’

3. Sketch the analysis for the following examples. Use the valence featuresSPR and COMPS rather than the abbreviations S, VP, and V′. Since thevalue of SPR in German is always the empty list, you may omit it inthe German examples. NPs and PPs can be abbreviated as NP and PP,respectively. Verb movement should be indicated with the ‘//’ symbol.

(38) a. dassthat

sieshe

darüberthere.upon

nachdenktPART.thinks

(German)

‘that she laughs about this’b. dass

thatsieshe

darüberthere.upon

nachdenkenPART.think

wirdwill

‘that she will think about this’


6.2 The analysis

c. Wirdwill

sieshe

darüberthere.upon

nachdenken?think



BjarneBjarne

ihærdigtseriously

påat

bogen?book.DEF

(Danish)


worksBjarneBjarne

ernsthaftseriously

anat

demthe

Buch?book

(German)

‘Does Bjarne work seriously on the book?’

4. Sketch the analysis of the following examples. NPs may be abbreviated.Valence features should not be given, but node labels like V, V′, VPand S should be used instead. If non-local dependencies are involvedindicate them using the ‘/’ symbol.

(40) a. Such books, I like.b. Solche

suchBücherbooks

maglike

ich.I

(German)

‘I like such books.’c. Boger

bookssomlike

detthis

elskerlike

jeg.I

(Danish)

‘I like such books.’


7 Passive

This chapter deals with the passive.1 Passive is usually analyzed as the suppres-sion of the subject. However, before we can develop an analysis we have to askwhat it is that constitutes a subject. This is a question that is the topic of editedvolumes and dissertations and modest as I am, I will try and provide an answerat least for the Germanic languages. As we will see, the situation is rather clearin languages like Danish, English, and German, but there are exciting facts to bediscovered about Icelandic.

7.1 The phenomenon

7.1.1 Subjects and other subjects

The situation in languages like English, Danish, and German is rather clear. Forinstance, many authors assume that non-predicative NPs in the nominative aresubjects in German. So, der Delphin ‘the dolphin’ is the subject of the sentencesin (1):

(1) a. Derthe

Delphindolphin

lacht.laughs

(German)

b. Derthe

Delphindolphin

hilfthelps

demthe

Kind.child

c. Derthe

Delphindoplhin

gibtgives

ihrher

einena

Ball.ball

The restriction to non-predicative NPs is needed since otherwise, we would haveto assume that both NPs in (2) are subjects, but ein Lügner ‘a lier’ is a predicativephrase and only der Mann ‘the man’ is the subject.

(2) Derthe.NOM

Mannman

istis

eina.NOM

Lügner.liar

(German)

‘The man is a liar.’

1This chapter is a rough draft. Most of the references to other literature are still missing …

7 Passive

In addition certain clausal arguments are treated as subjects.Genetives and datives as in (3) are not counted among the subjects in German.

(3) a. Ihrerthey.GEN

wurdewas

gedacht.remembered

(German)

‘They were remembered.’b. Ihm

he.DATwurdewas

geholfen.helped

‘He/she was helped.’

Interestingly the question whether genitives and datives like those in (3) aresubjects was answered quite differently for the SVO language Icelandic by re-searchers following the work of Zaenen, Maling & Thráinsson (1985). Althoughthe sentences in (4) look like those in (2), the genetive and the dative element areassumed to be subjects.

(4) a. Hennarshe.SG.GEN

varwas

saknað.missed

(Icelandic)

b. Þeimthey.PL.DAT

varwas

hjalpað.helped

Since Icelandic is a V2 language the constituent order in such simple sentencesdoes not help us to determine whether hennar ‘her’ and Þeim ‘them’ are subjectsor not. These elements are fronted and since both subjects and objects can befronted, the sentences in (4) do not help us in determining the grammatical func-tion of these arguments. However, Zaenen, Maling & Thráinsson (1985) arguedthat these elements should be analyzed as subjects and provided several tests.Among the tests are more elaborate positional tests and omissability in so-calledcontrol constructions. I will turn to these tests now.

7.1.1.1 The position of subjects in V2 and V1 sentences

The first test that was suggested uses the position of constituents in V2 sen-tences in which a non-subject is fronted (Zaenen, Maling & Thráinsson 1985:Section 2.3). For instance, consider the following examples:

(5) a. Meðwith

þessarithis

byssushotgun

skautshot

ÓlafurOlaf.NOM

refinn.the.fox.ACC

(Icelandic)

b. * Meðwith

þessarithis

byssushotgun

skautshot

refinnthe.fox.ACC

Ólafur.Olaf.NOM


7.1 The phenomenon

The nominative can appear directly after the finite verb skaut ‘shot’ as in (5a) butit cannot appear to the right of the accusative as in (5b).

The same can be observed with w-questions:

(6) a. Hvenærwhen

hafðihas

SiggaSigga.NOM

hjálpaðhelped

Haraldi?Harold.DAT

(Icelandic)

b. * Hvenærwhen

hafðihas

HaraldiHarald.DAT

SiggaSigga.NOM

hjálpað?helped

The object has to follow the participle hjálpað as in (6a) and the subject imme-diately follows the finite verb. Examples with the object before the subject as in(6b) are ungrammatical. The dative object can be fronted, but then it has to berealized in initial position to the left of the finite verb, not to its right:

(7) HaraldiHarald.DAT

hafðihas

SiggaSigga.NOM

aldreinever

hjálpað.helped

(Icelandic)

The same situation can be found in yes/no questions:

(8) a. Hafðihas

SiggaSigga.NOM

aldreinever

hjálpaðhelped

Haraldi?Harald.DAT

(Icelandic)

b. * Hafðihas

HaraldiHarald.DAT

SiggaSigga.NOM

aldreinever

hjálpað?helped

Zaenen, Maling & Thráinsson (1985: Section 2.3) observed that certain dativescan appear in this postverbal position as well:

(9) a. Hefurhas

hennishe.DAT

alltafalways

þóttthought

ÓlafurOlaf.NOM

leiðinlegur?boring.NOM

(Icelandic)

‘Has she always considered Olaf boring?’b. Ólafur

Olaf.NOMhefurhas

hennishe.DAT

alltafalways

þóttthought

leiðinlegur.boring.NOM

‘She alway considered Olaf boring.’c. * Hefur

hasÓlafurOlaf.NOM

henniher.DAT

alltafalways

þóttthought

leiðinlegur?boring.NOM

The German eqivalent would be the sentence in (10):

(10) ?? MichI.ACC

dünktthinks

derthe.NOM

Mannman

langweilig.boring

(German)

‘I think the man is boring.’


7 Passive

However, dünkt is archaic and is usually used with a dass clause – if it is used atall. But there is a non-archaic verb that has a similar form:

(11) MirI.DAT

scheintseems

derthe.NOM

Mannman

langweilig.boring

(German)

‘The man seems boring to me.’

The experiencer of scheinen ‘to seem’ is expressed with the dative, while thesubject of the embedded predicate langweilig ‘boring’ is in the nominative.

7.1.1.2 Subjects in control constructions

Zaenen, Maling & Thráinsson (1985: Section 2.7) discuss control structures inwhich the subject of the embedded verb is not expressed. (12a) shows an exampleof normal control inwhich the subject of thematrix verb vonast ‘to hope’ refers tothe same discourse referent as the subject of the embedded verb fara ‘to go’. (12b)is an example of so-called arbitrary control. In cases of arbitrary control there isno element depending on the head that governs the infinitive that refers to thesame discourse referent as the subject of the infinitive. The unexpressed subjectcorresponds to a pronoun one that is used generically. In example (22b) óvenjulegt‘unusual’ does not select for an argument that refers to the same referent as thesubject of fara ‘to go’. The subject of að fara heim snemma ‘to go home early’ isnot expressed but is understood as the indefinite pronoun one.

(12) a. ÉgI

vonasthope

tilfor

aðto

farago

heim.home

(Icelandic)

‘I hope to go home.’b. Að

tofarago

heimhome

snemmaearly

eris

óvenjulegt.unusual

‘It is unusual to go home early.’

Now, it can be observed that Icelandic allows verbs that do not take a nomi-native in such control constructions. An example is vantar (‘lacks’), which takestwo accusatives rather than a nominative and an accusative:

(13) MigI.ACC

vantarlack

peninga.money.ACC

(Icelandic)

(14) shows that this verb can be embedded under vonast (‘to hope’):

(14) ÉgI

vonasthope

tilfor

abto

vantalack

ekkinot

peninga.money.ACC

(Icelandic)

‘I hope that I do not lack money.’


7.1 The phenomenon

This should be compared with German:

(15) a. MirI.DAT

fehltlacks

keinno.a.NOM

Geld.money

(German)

‘I do not lack money.’b. * Ich

Ihoffe,hope

keinnot.a.NOM

Geldmoney

zuto

fehlen.lack

Intended: ‘I hope that I do not lack money.’

The question at the beginning of this section was whether the datives andgenetives in sentences like (4) – repeated here as (16) – are subjects or not.

(16) a. Hennarshe.GEN

varwas

saknað.missed

(Icelandic)

‘She was missed.’b. Þeim

they.DATvarwas

hjalpað.helped

‘They were helped.’

We are now able to use the tests to answer this question: the dative is right-adjacent to the finite verb in the question in (17) and hence in subject position.

(17) a. Varwas

hennarshe.GEN

saknað?missed

(Icelandic)

‘Was she missed?’b. Var

wasÞeimthey.DAT

hjálpað?helped

‘Were they helped?’

Similarly the dative follows the finite verb in the V2 sentence in (18):

(18) Íin

prófinuthe.exam

varwas

honumhe.DAT

vistapparently

hjálpað.helped

(Icelandic)

‘Apparently he was helped in the exam.’

In addition these datives can be omitted in control constructions as the exam-ples in (19) show:

(19) a. ÉgI

vonasthope

tilfor

aðto

verðabe

hjálpað.helped

(Icelandic)


7 Passive

b. Aðto

verabe

hjálpaðhelped

iin

prófinuthe.exam

eris

óleyfilegt.un.allowed

‘It is not allowed to be helped in the exam.’

This should be compared to German: While verbs like unterstützen ‘to support’that govern a nominative and an accusative can appear in such control construc-tions, verbs like helfen ‘to help’ that take a nominative and a dative are ruled outin this construction:

(20) a. dassthat

jemandsomebody

ihmhim.DAT

hilfthelps

(German)

b. dassthat

jemandsomebody

ihnhim.ACC

unterstütztsupports

c. dassthat

ihmhim.DAT

geholfenhelped

wirdis

d. dassthat

erhe.NOM

unterstütztsupported

wirdis

(21) a. IchI

hoffehope


zuto

werden.be

(German)

b. * IchI

hoffehope

geholfenhelped

zuto

werden.be

The dative object cannot be omitted in such control constructions as (21b) shows.The only way to realize a passive below hoffen is to use the dative passive witherhalten/bekommen/kriegen. The dative passive can turn a dative object into anominative subject:

(22) AickeAicke.NOM

bekommtgets

geholfen.helped

(German)

‘Aicke gets helped.’

Since the object of helfen is then nominative and hence undoubtfully a subjectin German, it does not come with a surprise that it can be omitted in controlconstructions like (23):

(23) IchI

hoffehope

hierhere

geholfenhelped

zuto

bekommen.2

get(German)

‘I hope to get help here.’2http://www.photovoltaikforum.com/sds-allgemein-ueber-solar-log-f38/solarlog-1000-mit-wifi-anschliesen-t96371.html. 10.01.2014


http://www.photovoltaikforum.com/sds-allgemein-ueber-solar-log-f38/solarlog-1000-mit-wifi-anschliesen-t96371.html

http://www.photovoltaikforum.com/sds-allgemein-ueber-solar-log-f38/solarlog-1000-mit-wifi-anschliesen-t96371.html

7.1 The phenomenon

7.1.2 Comparison German, Danish, English, Icelandic

In the following subsections I want to compare several dimensions in which theGermanic languages vary:

• Danish and Icelandic have a morphological passive, English and Germando not.

• German, Icelandic allow for subjectless constructions, Danish and Englishdo not.

• Danish, German and Icelandic allow for impersonal passives, English doesnot.

• Danish and Icelandic allow both objects to be promoted to subject, Englishand German do not.

• German has the so-called remote passive, Danish the so-called complexpassive and Danish and English have the so-called reportive passive.

7.1.2.1 Morphological and Analytic Forms

Danish allows for a morphological passive. It is formed by appending the -s suffixand there are present tense (24b) and past forms (24c):

(24) a. PeterPeter

læserreads

avisen.newspaper.DEF

(Danish)

‘Peter is reading the newspaper.’b. Avisen

newspaper.DEFlæsesread.PRES.PASS

afby

Peter.Peter

‘The newspaper is read by Peter.’c. Avisen

newspaper.DEFlæstesread.PAST.PASS

afby

Peter.Peter

‘The newspaper was read by Peter.’

Danish also allows for the analytic form with blive ‘be’ and participle:

(25) Avisennewspaper.DEF

bliveris

læstread

afby

Peter.Peter

(Danish)

‘The newspaper is read by Peter.’

The morphological passive may also apply to infinitives:


7 Passive

(26) Avisennewspaper.def

skalmust

læsesread.INF.PASS

hverevery

dag.day

(Danish)

‘The newspaper must be read every day.’

English and German only have the analytic variant:

(27) a. The paper was read.b. Der

the.NOMAufsatzpaper

wurdewas

gelesen.read

(German)

7.1.2.2 Personal and impersonal Passive

All languages under consideration allow for the promotion of an accusative ob-ject to subject. As the following examples show, the subject can be an S or aVP:

(28) a. Atthat

regeringengovernment.DEF

træderresigns

tilbage,PART

bliveris

påstået.claimed

(Danish)

‘It is claimed that the government resigns.’b. At

toreparererepair

bilen,car.DEF

bliveris

forsøgt.tried

‘It is tried to repair the car.’

In addition to such personal passives, Danish, German, and Icelandic allowfor impersonal passives. Since German does not require a subject, impersonalpassives like (29) are expected:

(29) weilbecause

nochstill

getanztdanced

wurdewas

(German)

‘because there was still dancing there’

The following two examples from Icelandic show that Icelandic also knowsimpersonal constructions (Thráinsson 2007: 264):

(30) a. Oftoften

varwas

talaðtalked

umabout

þennanthis

mann.Mann.ACC.SG.M

(Icelandic)

b. Aldreinever

hefurhas

veriðbeen

sofiðslept

íin

þessuthis

rúmi.bed.DAT

‘This bed has never been slept in.’


7.1 The phenomenon

Danish also allows for impersonal passives but it differs from the languagesdiscussed so far in that it requires an expletive subject:

(31) a. atthat

derEXPL

bliveris

dansetdanced

(Danish)

‘that there is dancing’b. at

thatderEXPL

dansesdance.PRES.PASS

‘that there is dancing’c. * Bliver

isdanset?danced

d. * Danses?dance.PASS

So Danish is like English in always requiring a subject, but while this constraintresults in the impossibility of impersonal passives in English, Danish found asolution to the subject problem by inserting an expletive.

Expletives are excluded in German impersonal constructions:

(32) * weilbecause

esit

nochstill

gearbeitetworked

wurdewas

(German)

Intended: ‘because there was still working there’

7.1.2.3 Promotion of the primary and secondary object

English and German allow the promotion of one of the objects of a ditransitiveverb only. (33) shows that the accusative object can be realized as subject, butthe dative cannot:

(33) a. weilbecause

derthe.NOM

Mannman

demthe.DAT

Jungenboy

denthe.ACC

Ballball

schenktgives

(German)

‘because the man gives the boy a ball as a present’b. weil

becausedemthe.DAT

Jungenboy

derthe.NOM

Ballball

geschenktgiven

wurdewas

‘because the ball was given to the boy’c. * weil

becausederthe.NOM

Jungeboy

denthe.ACC

Ballball

geschenktgiven

wurdewas


7 Passive

Similarly, English can realize the first object as subject, but the second objectcannot be promoted to subject:

(34) a. because the man gave the boy the ballb. because the boy was given the ballc. * because the ball was given the boy

The information structural effect can be reached with a different lexical variantof give though. give can be used with an NP object and a to PP instead of twoNPs as in (35a). The first object of the ditransitive give is realized as PP in (35a)and the second object the ball is the first object in (35a). This alternation is alsocalled dative-shift.

(35) a. because the man gave the ball to the boyb. because the ball was given to the boy

(36b) is the passive variant of (35a). As in (34b), the primary object is promotedto subject.

Danish and Icelandic differ from English and German. In the former languagesboth objects can be promoted to subject without any previous alternation of va-lence frames like dative shift.


mandenman.DEF

givergives

drengenboy.DEF

boldenball.DEF

(Danish)

‘because the man gives the boy the ball’b. fordi

becausedrengenboy.DEF

bliveris

givetgiven

boldenball.DEF

‘because the boy is given the ball’c. fordi

becauseboldenball.DEF

bliveris

givetgiven

drengenboy.DEF

‘because the ball is given to the boy’

One could assume that it is always the first object (the primary object) that ispromoted to subject and that Danish does not have an order of the objects, sothat both objects are equally prominent and can be promoted to subject. Moro isa language that is said to have such properties (Ackerman et al. 2013). However,Danish differs from Moro in that the order of the objects in sentences is clearlyfixed: While (36a) is possible, the reverse order of the objects is ungrammaticalas (37) shows.

(37) * fordi manden giver bolden drengen


7.1 The phenomenon

As far as Icelandic is concerned, Zaenen, Maling & Thráinsson (1985: 460) notethat apart from the possibility to promote the accusative to subject, the dativecan become a quirky subject:

(38) Konunginumthe.king.DAT

voruwere

gefnargiven.F.PL

ambáttir.maidservants.NOM.F.PL

‘The king was given female slaves.’

The structure of (38) is sketched in (39):

(39) [S𝑖 Aux _𝑖 V O]

Since the nominative is serialized after the participle it cannot be the subject,which implies that the fronted dative element is the subject.

Alternatively the accusative object is promoted to subject:

(40) Ambáttinthe.maidservant.NOM.SG

varwas

gefingiven.F.SG

konunginum.the.king.DAT

‘The female slave was given to the king.’

This sentence also has the structure in (39).In order to show that the dative is really promoted to subject in (38) and the

accusative is promoted to subject in (40), Zaenen, Maling & Thráinsson (1985:460) apply a battery of tests. I only give the V2 examples with an adjunct ininitial position, the questions, and the control structures here. The examples in(41) and (42) show that the sentences above really have the structure in (39). Thefirst position in (41) is filled by an adjunct, which entails that the subject remainsin subject position and hence shows that the dative konunginum ‘the king’ is thesubject. Similarly the nominative ambfittin is the subject in (40b).

(41) a. Umin

veturinnthe.winter

voruwere

konunginumthe.king.DAT

gefnargiven

ambfittir.slaves.NOM

‘In the winter, the king was given (female) slaves.’b. Um

inveturinnthe.winter

varwas

ambfittinthe.slave.NOM

gefingiven

konunginum.the.king.NOM

‘In the winter, the slave was given to the king.’

The questions in (42) are further evidence. The initial position is not filled andthe dative in (42a) and the nominative in (42b) is realized prenominally.

(42) a. Voruwere


gefnargiven

ambfittir?slaves.NOM

‘Was the king given slaves?’


7 Passive

b. Varwas

ambfittinthe.slave.NOM

gefingiven

konunginum?the.king.DAT

‘Was the slave given to the king?’

(43) shows the respective control examples:

(43) a. Aðto

verabe

gefnargiven

ambáittirslaves.NOM

varwas

mikillgreat

heiður.honor

‘To be given slaves was a great honor.’b. Að

toverabe

gefingiven


ollicaused

miklumgreat

vonbrigðum.disappointment

‘To be given to the king caused great disappointment.’

In (43a) the dative is not expressed and in (43b) the nominative is omitted. Thisshows that both the primary and the secondary object can be promoted to subjectin Icelandic, even though one of them has structural and the other one lexicalcase.

7.2 The Analysis

7.2.1 Structural and lexical case and the Case Principle

In order to analyze the passive it is useful to distinguish between structural andlexical case. Structural case is case that depends on the syntactic structure inwhich arguments get realized, while lexical case is case that stays constant inde-pendent of the syntactic environment. In addition to lexical and structural casethere is semantic case. This case is not assigned by a governing head like a verb,adjective or preposition but is due to a certain function of an adverbial. For in-stance time expressions like den ganzen Tag ‘the whole day’ in (44) are in theaccusative in German.

(44) Erhe.NOM

arbeitetworks

denthe.ACC

ganzenwhole

Tag.day

(German)

‘He works the whole day.’

Since this chapter is about the passive and its variation in the Germanic lan-guages, I will ignore semantic case here.

7.2.1.0.1 Nominatives and accusative objects

Until now the case that an argument gets assigned by its head was representedin the valency list of the head. With such a representation we would need two


7.2 The Analysis

different lexical items for the verb lesen (‘to read’): one in which the verb takesa nominative and an accusative as in (45c) and one in which it takes two ac-cusatives as in (45d).

(45) a. Erhe.NOM

wirdwill

dasthe.ACC

Buchbook

lesen.read

(German)

‘He will read the book.’b. Ich

Isahsaw

ihnhim

dasthe

Buchbook

lesen.read

‘I saw him read the book.’c. ⟨ NP[nom], NP[acc] ⟩d. ⟨ NP[acc], NP[acc] ⟩

Rather than having these two lexical items, one can have just one lexical item andleave the actual case assignment for later. So depending on whether the subjectof lesen is realized as the subject of wird or as the object of sah ‘saw’, it getsnominative or accusative. Such cases are called structural cases. The distinctionbetween structural and lexical case will play an important role in the analysisof passive. It is this distinction that makes a unified analysis of the so-calledpersonal and impersonal passive possible.

(46) provides additional examples and involves different forms of the verb (fi-nite vs. non-finite) and a nominalization:

(46) a. Derthe

Installateurplumber

kommt.comes

‘The plumber comes.’b. Der

theMannman

läßtlets

denthe

Installateurplumber

kommen.come

‘The man lets the plumber come.’c. das

theKommencoming

desof.the

Installateursplumber

‘the coming of the plumber’

The example in (46c) also shows that the subject of kommen ‘to come’ can be re-alized as genetive. So, nominative, genitive, and accusative are structural casesin German. (The question whether some or all datives should be treated as struc-tural case is addressed below).

The examples in (46) show that the case of subjects in German can change,those in (47) show that the case of accusative objects can change as well:


7 Passive

(47) a. JuditJudit

schlägtdefeats

denthe.ACC

Weltmeister.world.champion

‘Judit defeats the world champion.’b. Der

the.NOMWeltmeisterworld.champion

wirdis

geschlagen.beaten

‘The world champion is beaten.’

7.2.1.0.2 Genitive objects

The examples in (48) show instances of lexical case: Genitive that depends onthe verb is lexical since it does not change when the verb is passivized.

(48) a. Wirwe.NOM

gedenkenremember

derthe

Opfer.victims.GEN

‘We remember the victims.’b. Der

the.GENOpfervictims

wirdis

gedacht.remembered

‘The victims are remembered.’c. * Die

the.NOMOpfervictims

wirdis

/ werdenare

gedacht.remembered

As the example in (48c) shows, the nominative is impossible. The genitive objectremains in the genitive in passive constructions. Passives without a subject as in(48b) are traditionally called impersonal passives.

7.2.1.0.3 Dative objects

Now, lets turn to the dative. If we look at examples like (49), we see that thedative does not change either in the passive:

(49) a. DertheNOM

Mannman

hathas

ihmhim.DAT

geholfen.helped

‘The man helped him.’b. Ihm

him.DATwirdis

geholfen.helped

‘He is helped.’

So in analogy to the genitive examples above, the dative should be a lexical case.


7.2 The Analysis

But there are examples like those in (50) and according to the view that struc-tural cases are those cases that vary according to the syntactic environment, thedative should be a structural case.

(50) a. Derthe

Mannman

hathas

denthe

Ballball

demthe

Jungenboy

geschenkt.given

‘The man gave the boy the ball as a present.’b. Der

theJungeboy

bekamgot

denthe

Ballball

geschenkt.given

‘The boy got the ball as a present.’

The question whether the dative should be seen as a structural or a lexical case isa hotly debated one. In principle there are three possibilities and all three of themwere suggested in the literature. One could assume that all datives are lexical,that some are lexical and others are structural, or that all datives are structural.

I follow Haider (1986a) and treat all datives as lexical cases. Under this as-sumption, the contrast in Haider’s examples (1986a: 20) in (51) is explained im-mediately:

(51) a. Erhe.NOM

streicheltstrokes

denthe.ACC

Hund.dog

b. Derthe.NOM

Hunddog

wurdewas

gestreichelt.strocked

c. seinhis

Streichelnstrocking

desof.theġen

Hundesdog

d. Erhe

hilfthelps

denthe.DAT

Kindern.children

e. Denthe.DAT

Kindernchildren

wurdewas

geholfen.helped

f. dasthe

Helfenhelping

derof.the

Kinderchildren

g. * seinhis

Helfenhelping

derthe

Kinderchildren

The accusative object of streicheln ‘to stroke’ can be realized as nominative in thepassive, so it is clearly a structural case. Nominalizations allow this object to berealized in the genitive as (51c) shows. However, this does not work with datives.The dative object of helfen ‘to help’ cannot be realized in the genitive. (51f) is


7 Passive

possible, but only with a reading in which the children are the agents, that is, thenominalization in (51f) corresponds to (52) rather than (51d):

(52) Diethe.NOM

Kinderchildren

helfenhelp

jemandem.somebody

If the agent is expressed by a prenominal possessive as in (51g) the genetive ordative der Kinder is ruled out.

The only way to express the dative at all is prenominally:

(53) dasthe

Den-Kindern-Helfenthe-children-helping

‘the children’s helping’

So, authors who assume that all datives are structural have a problem explain-ing the differences in impersonal passives and nominalizations. In addition thereis a problemwith bivalent verbs. While some verbs take the dative others take theaccusative although there is hardly any semantic difference or any other reasonthat could be made responsible.

(54) a. Erhe

hilfthelps

ihm.him

b. Erhe

unterstütztsupports

ihn.him

The fact that helfen takes a dative object, while unterstützen takes an accusativeis just an idiosyncrasy of German that speakers of German have to learn whenthey acquire the language. So, the information in the lexical entry for helfenmust be different from the one for unterstützen. Some authors acknowledge thisdifference and assume that the dative of bivalent verbs is lexical, while the da-tive of ditransitive verbs is structural. The assumption is that verbs assign thenominative to their first argument, the accusative to their last argument and ifthere is an additional argument that is neither the first nor the last, it gets dative.The prediction that such mixed accounts make is that the dative passive shouldbe possible with ditransitive verbs but impossible with bivalent verbs, since thedative is structural for the former verbs and lexical for the latter. The empiricalsituation is not as clear-cut as one might wish. Some authors accept exampleslike (55). Others reject them.

(55) a. Erhe

kriegtegot

vonby

vielenmany

geholfenhelped

/ gratuliertcongratulated

/ applaudiert.applauded


7.2 The Analysis

b. Manone

kriegtgets

täglichdaily

gedankt.thanked

However, there are attested examples:

(56) a. „Da kriege ich geholfen.“3

b. Heute morgen bekam ich sogar schon gratuliert.4

c. „Klärle“ hätte es wirklich mehr als verdient, auch mal zu einem„unrunden“ Geburtstag gratuliert zu bekommen.5

d. Mit dem alten Titel von Elvis Presley „I can’t help falling in love“bekam Kassier Markus Reiß zum Geburtstag gratuliert, […]6

I think that the verbs kriegen, erhalten, and bekommen are on the way to becomeauxiliaries. Their meaning is getting more and more bleached. Hence there arealmost no selectional restrictions left on the downstairs verb. The only require-ment for the dative passive to apply is of course that the embedded verb governsa dative.

Now, if the dative passive is possible with verbs bivalent like helfen and ifhelfen has to govern a lexical dative (since otherwise the difference betweenhelfen and unterstützen could not be explained), it follows that the dative pas-sive must be able to convert a lexical dative into a structural case (realized asnominative in the examples above). This means that one could assume that alldatives are lexical, even the datives of ditransitive verbs. This explains why thesedatives are not realized as nominatives or accusatives in passives like (57):

(57) a. dassthat

erhe.NOM

demthe.DAT

Jungenboy

denthe.ACC

Ballball

gegebengiven

hathas

b. dassthat

demthe.DAT

Jungenboy

derthe.NOM

Ballball

gegebengiven

wurdewas

c. * dassthat

derthe.NOM

Jungeboy

denthe.ACC

Ballball

gegebengiven

wurdewas

d. * dassthat

denthe.ACC

Jungeboy

derthe.NOM

Ballball

gegebengiven

wurdewas

They just stay dative. The only exception is the dative passive and this has to beanalyzed as an exception.

3Frankfurter Rundschau, 26.06.1998, p. 7.4Brief von Irene G. an Ernst G. vom 10.04.1943, Feldpost-Archive mkb-fp-02705Mannheimer Morgen, 28.07.1999, Lokales; „Klärle“ feiert heute Geburtstag.6Mannheimer Morgen, 21.04.1999, Lokales; Motor des gesellschaftlichen Lebens.


7 Passive

After this discussion of lexical and structural case in German, I will providethe Case Principle, which is responsible for case assignment. As was explained inSection 4.8, it is assumed that all arguments of a head are represented in one list:the ARGUMENT STRUCTURE list (ARG-ST list). (58) shows the argument structurelist of a ditransitive verb like geben ‘to give’:

(58) ⟨ NP[str], NP[ldat], NP[str] ⟩

As was argued above, dative is treated as a lexical case. ldat is an abbreviationfor lexical dative and str stands for for structural case. The Case Principle hasthe following form (adapted from Przepiórkowski 1999; Meurers 1999):

Principle 1 (Case Principle)• In a list that contains both the subject and the complements of a verbalhead, the first element with structural case gets nominative unless it israised by a higher head.

• All other elements in this list that have structural case and are not raisedget accusative.

• In nominal environments elements with structural case get genitive.

This principle is inspired by Yip, Maling & Jackendoff (1987) and as will be demon-strated below it also works for the complex case system in Icelandic. It differsfor in not assigning case to elements that are raised to a higher predicate. Thispoint will be explained in more detail below.

The effect of this principle will be explained with respect to the verbs in (59):

(59) a. schläft ‘sleep’: ARG-ST⟨NP[str]𝑖

⟩b. unterstützt ‘support’: ARG-ST

⟨NP[str]𝑖 , NP[str]𝑗

⟩c. hilft ‘help’: ARG-ST

⟨NP[str]𝑖 , NP[ldat]𝑗

⟩d. schenkt ‘give as a present’: ARG-ST

⟨NP[str]𝑖 , NP[ldat]𝑗 , NP[str]𝑘

⟩The first element in these lists that has structural case gets nominative and thesecond one accusative. This is exactly what one expects. The result is given in(60). snom stands for structural nominative.


7.2 The Analysis

(60) a. schläft ‘sleep’: ARG-ST⟨NP[snom]𝑖

⟩b. unterstützt ‘support’: ARG-ST

⟨NP[snom]𝑖 , NP[sacc]𝑗

⟩c. hilft ‘help’: ARG-ST

⟨NP[snom]𝑖 , NP[ldat]𝑗

⟩d. schenkt ‘give as a present’: ARG-ST

⟨NP[snom]𝑖 , NP[ldat]𝑗 , NP[sacc]𝑘

⟩7.2.2 Argument reduction and case assignment: The passive

Now, with the structural/lexical case distinction the analysis of the passive isreally simple and directly corresponds to the intuition that the passive is thesuppression of the subject (the most prominent, that is, the first argument in theARG-ST list). If the first argument is removed from the lists in (59), the followinglists result:

(61) a. geschlafen: ARG-ST ⟨ ⟩

b. unterstützt: ARG-ST⟨NP[str]𝑗

⟩c. geholfen: ARG-ST

⟨NP[ldat]𝑗

⟩d. geschenkt: ARG-ST

⟨NP[ldat]𝑗 , NP[str]𝑘

⟩The NPs that are in the first position in (61) where in the second position in (60).The first NP with structural case gets nominative and hence the following caseassignments result:

(62) a. geschlafen: ARG-ST ⟨ ⟩

b. unterstützt: ARG-ST⟨NP[snom]𝑗

⟩c. geholfen: ARG-ST

⟨NP[ldat]𝑗

⟩d. geschenkt: ARG-ST

⟨NP[ldat]𝑗 , NP[snom]𝑘

⟩Lexical case as in (62c–d) is not affected by the case principle, it stays the way itwas specified, namely dative.

It should be noted here that this simple approach to passive accounts bothfor the so-called personal and the impersonal passive. The passives of schlafen‘to sleep’ and helfen ‘to help’ are called impersonal passives since the respectiveclauses do not have a subject.


7 Passive

(63) a. dassthat

geschlafenslept

wurdewas

‘that there was sleeping there’b. dass

thatdemthe.DAT

Mannman

geholfenhelped

wurdewas

‘that the man was helped’

The passives of unterstützen ‘to support’ and schenken ‘to give as a present’ dohave subjects, namely the arguments that are realized as accusative objects inthe active:

(64) a. dassthat

derthe.NOM

Mannman


wurdewas

‘that the man was supported’b. dass

thatdemthe.DAT

Jungenboy

dertheNOM

Ballball

geschenktgiven

wurdewas

‘that the ball was given to the boy as a present’

Those analyses that assign all cases lexically would have to assume that the caseof the objects (accusative) is changed into nominative in the passive. Hence therewould be two variants of the passive: The impersonal passive just suppresses thesubject and the personal one suppresses the subject and additionally changes thecase of the object into nominative. The analysis using the structural/lexical casedistinction just postpones the case assignment until the point where it is clearwhat the right case will be. If we have a participle and use it with the passiveauxiliary it is clear what the case of the arguments has to be.

7.2.3 Argument extension and case assignment: AcI constructions

The case principle contains restrictions on case assignment that prohibits the as-signment to elements that are raised. These restrictions have not been explainedyet. Consider the examples in (65):

(65) a. Derthe.NOM

Jungeboy

liestreads

denthe.ACC

Aufsatz.paper

‘The boy reads the paper.’b. Der

the.NOMMannman

läßtlets

denthe.ACC

Jungenboy

denthe.ACC

Aufsatzpaper

lesen.read

‘The man lets the boy read the paper.’


7.2 The Analysis

The example (65a) shows that the subject of lesen is assigned nominative. How-ever, the subject of lesen gets accusative in (65b). So, if one would assign caseon the basis of the argument structure of lesen in (65b), one would assign nomi-native, but the AcI verb lassen ‘to let’ assigns accusative to its object. The pointis that the subject of lesen is raised to the object of lassen. The Case Principle isset up in a way such that case is assigned only to those arguments that are notraised to a higher head. Hence, den Jungen does not get case from lesen, but fromlässt.

The analysis of (65b) is given in Figure 7.1. The arguments of lesen ‘to read’

V[⟨ ⟩]

NP[snom]

der Mannthe man

V[⟨ NP[snom] ⟩]

NP[sacc]

den Jungenthe boy

V[⟨ NP[snom], NP[sacc]⟩]

NP[sacc]

den Aufsatzthe paper

V[⟨ NP[snom], NP[sacc], NP[sacc]⟩]

V[⟨ NP[sacc], NP[sacc]⟩]

lesenread

V[⟨ NP[snom], NP[sacc], NP[sacc], V ⟩]

lässtlets

Figure 7.1: Analysis of AcI constructions as raising constructions and the verbalcomplex in German

are taken over by lässt. Since lässt contributes its own argument, the causer orthe one who gives the permission, lässt selects for three NPs with structural caseand a verb in the specific sentence depicted in Figure 7.1. According to the CasePrinciple the first NP with structural case gets nominative and the other NPswith structural case get accusative. This results into a list with one NP in thenominative and two NPs in the accusative.

(66) shows the ARG-ST list of lässt when it is combined with schlafen, unter-stützen, helfen, or schenken, respectively.

(66)


7 Passive

a. läßt with schlafen: ARG-ST⟨NP[str]𝑙 , NP[str]𝑖 , V

⟩b. läßt with unterstützen : ARG-ST

⟨NP[str]𝑙 , NP[str]𝑖 , NP[str]𝑗 , V

⟩c. läßt with helfen: ARG-ST

⟨NP[str]𝑙 , NP[str]𝑖 , NP[ldat]𝑗 , V

⟩d. läßt with schenken: ARG-ST

⟨NP[str]𝑙 , NP[str]𝑖 , NP[ldat]𝑗 , NP[str]𝑘 , V

⟩The NP that is added has the index l. As the first NP with structural case on theselists it gets nominative. All other elements of this list that have structural caseget accusative. Hence the subject of the embedded verb is assigned accusative,the lexical cases stay the same and the accusative objects of the embedded verbget accusative as well, since their case is structural too.

Note that the question of whether a language has a verbal complex or not isorthogonal to issues of case assignment. Figure 7.2 shows the analysis of the En-glish translation of (65b). let selects for the subject, the object and a VP. The sub-

S

NP[snom]

the man

VP

V

V

lets

NP[sacc]

the boy

VP

V

read

NP[sacc]

the book

Figure 7.2: AcI constructions in English

ject of read is simultaneously the object of let and hence the Case Principle doesnot assign nominative to the subject of the embedded verb read, but accusativeto the object of the matrix verb let.

7.2.4 Theoretical analysis of the crosslinguistic differences

Argument reduction and case assignment was already explained for German inSection 7.2.2. I want to get a little bit more explicit now and provide lexical itemsfor the passive and perfect auxiliary for German. After this I discuss the other


7.2 The Analysis

languages and explain how the differences can be dealt with in a worked outanalysis.

7.2.4.1 Designated Argument Reduction

Haider (1986a) suggested marking the argument of a verb that has subject prop-erties. He calls these special arguments designated argument. Heinz & Matiasek(1994) transferered this idea to HPSG and Müller (2003b) modified it slightly toget certain facts with modal infinitives right. One important use of the desig-nated argument is to distinguish so-called unaccusative verbs from unergativeverbs. Perlmutter (1978) pointed out that unaccusative verbs have remarkableproperties and argued that their subjects are not really subjects but behave morelike objects. One of their properties is that they do not allow for passives. Fur-thermore their participles can be used attributively which is usually not possible:

(67) a. derthe

angekommenearrived

Zugtrain

‘the arrived train’b. * der

thegeschlafeneslept

Mannman

This is explained if one assumes that the subject of ankommen ‘arrive’ is indeedlike an object. As an object it patterns with the object of transitive verbs:

(68) derthe

geliebtebeloved

Mannman

Mann ‘man’ fills the object slot of geliebte. If the sole argument of ankommenis treated as an object, the similarity to the transitive lieben is explained im-medeately. Similarly the fact that unaccusatives do not allow for passives is ex-plained: If passive is the suppression of the subject and ankommen does not havea subject in that sense, passive cannot apply.

(69) a. Derthe

Zugtrain

istis

angekommen.arrived

‘The train arrived.’b. * weil

becauseangekommenarrived

wurdewas

In theHPSG analyses the authors assume that there is a list-valued feature DES-IGNATED ARGUMENT (DA). This list contains the subject of transitive and unerga-


7 Passive

tive verbs (intransitive verbs that are not unaccusative). The DA value of unac-cusative verbs is the empty list, since these verbs do not have an argument withsubject properties.

The passive is analyzed as a lexical rule that licences a lexical item for theparticiple. The ARG-ST list of the particple is the ARG-ST list of the verb stem thatis the input to the lexical rule minus the DA list. Since this is not the focus of thisbook, I will not discuss unaccusative verbs in the following. (70) provides someprototypical examples for unergative and transitive verbs:

(70) ARG-ST DA

a. tanzen (dance): ⟨ 1NP[str] ⟩ ⟨ 1 ⟩b. lesen (read): ⟨ 1NP[str], NP[str] ⟩ ⟨ 1 ⟩c. schenken (give as a present): ⟨ 1NP[str], NP[ldat], NP[str] ⟩ ⟨ 1 ⟩d. helfen (help): ⟨ 1NP[str], NP[ldat] ⟩ ⟨ 1 ⟩

The lexical rule that forms the participle is sketched in (71):

(71) Lexical rule for the formation of the participle (preliminary):stem

HEAD[verbDA 1

]ARG-ST 1 ⊕ 2

↦→

[wordARG-ST 2

]

This rule splits the ARG-ST list of the input into two lists 1 and 2 . 1 is identicalto the DA value. Therefore the designated argument is taken off the ARG-ST listand is not present in the lexical item that is licensed by the rule.

The ARG-ST list of the participle that is licensed is either empty (72a) or startswith an object of the active form:

(72) ARG-ST

a. getanzt (danced, unerg): ⟨⟩b. gelesen (read, trans):

⟨NP[str]

⟩c. geschenkt (given, ditrans):

⟨NP[ldat], NP[str]

⟩d. geholfen (helped, unerg):

⟨NP[ldat]

⟩As was explained above, the first element in the ARG-ST list with structural casegets nominative and hence the accusative object of lesen in (73a) is realized asnominative in (73b):


7.2 The Analysis

(73) a. Erhe.NOM

liestreads

denthe.ACC

Aufsatz.paper

b. Derthe.NOM

Aufsatzpaper

wurdewas

gelesen.read

English differs from German in not having a dative case at all. I am talkingabout morphological markings here, not about semantics. Therefore both objectsof English ditransitive verbs are accusative objects. However, only one of theobjects can be promoted to subject. This is modeled in the analysis at hand byassuming that the secondary object bears lexical accusative (see also Grewendorf(2002: 57) for the assumption of lexical accusative for the secondary object inEnglish).7

(74) ARG-ST

b. dance (unerg):⟨NP[str]

⟩c. read (trans):

⟨NP[str], NP[str]

⟩d. give (ditrans):

⟨NP[str], NP[str], NP[lacc]

⟩e. help (trans):

⟨NP[str], NP[str]

⟩German can promote the second object (accusative) and English the first one. Thecommonality is that the object that is closer to the verb can be promoted. This isthe accusative for German since nominative, dative, accusative is the unmarkedorder and German is a OV language and the first accusative in English sinceEnglish is a VO language.

(75) a. dassthat

demthe.DAT

Jungenboy

derthe.NOM

Ballball

gegebengiven

wurdewas

‘that the ball was given to the boy’b. because the boy was given the ball

A further difference is the lexical item for help. Since there is no dative in English,the object is marked accusative as it is the case for read. Interestingly, Englishallows for the personal passive of help, while this is not possible in German:

(76) a. because he was helped

7Admittidly this is just a restatement of the facts, since assigning lexical case means that theargument under consideration cannot have another case. But taken together with constraintson subjects in English the facts about promotion or non-promotion of arguments follow nicely.


7 Passive

b. weilbecause

ihmhe.DAT

geholfenhelped

wurdewas

c. * weilbecause

erhe.NOM

geholfenhelped

wurdewas

7.2.4.2 Primary and secondary objects

In this section I want to look at languages that allow both objects to be promoted.Danish is like English in not having a dative. This is refelcted in the followingARG-ST values:

(77) ARG-ST

a. danse (dance, unerg):⟨NP[str]𝑖

⟩b. læse (read, trans):


⟩c. give (give, ditrans):

⟨NP[str]𝑖 , NP[str]𝑗 , NP[str]𝑘

⟩d. hjælpe (help, trans):


⟩Danish has two objects with structural case, English and German have just oneobject with structural case and the other one with lexical accusative and lexicaldative, respectively. Since English and German do not allow for subjects withlexical case it is clear that the promotion to subject of the argument that bearslexical case is excluded. Danish also disallows subjects with lexical case, but sincethe two objects have structural case anyway, they both can be promoted.

Note however that the lexical rule in (71) does not account for the promotionof the secondary object. What it does is suppressing the subject. Under the as-sumption that the first NPwith structural case is the subject, the secondary objectcould never be realized as subject. Note that it would not help to say any NP withstructural case can be the subject, since this would admit wrong realizations. Inaddition to the correct (36a), the following two sentences would be admitted:

(78) a. * fordibecause

drengenboy.DEF

givergives

mandenman.DEF

boldenball.DEF

b. * fordibecause

boldenball.DEF

givergives

mandenman.DEF

drengenboy.DEF

(78a) is ungrammatical with drengen ‘boy’ the recipient of the giving. Similarlythe transfered object bolden cannot be realized as subject in active sentences.This means that the promotion to subject has to be a part of the lexical rule thatlicences the participle that is used in the passive. The lexical rule in (79) takes


7.2 The Analysis

the ARG-ST list and splits it into two lists. The first list 1 is identical to the valueof DA. The second list 2 is the remainder of the ARG-ST list. 2 is related to 3 bythe relational constrint promote. 3 is either eaqual to 2 or additional provides alist in which another NP with structural case is positioned at the beginning of 3 .

(79)Lexical rule for the passive for Danish, English, German, and Icelandic:HEAD

[verbDA 1

]ARG-ST 1 ⊕ 2

↦→[ARG-ST 3

]∧ promote( 2 , 3 )

(80) shows the ARG-ST values of our prototypical verbs:

(80) ARG-ST

a. danset/-s (dance, unerg): ⟨⟩b. læst/-s (read, trans):

⟨NP[str]𝑗

⟩c. givet/-s (give, ditrans):

⟨NP[str]𝑗 , NP[str]𝑘

⟩⟨NP[str]𝑘 , NP[str]𝑗

⟩d. hjulpet/-s (help, trans):

⟨NP[str]𝑗

⟩The NP[str]𝑖 that is the first element in (77) is suppressed. The effect of promoteis that there are two different ARG-ST values for the passive variants of givet ‘togive’: one with an ARG-ST list in which NP[str]𝑗 preceedes NP[str]𝑘 and anotherone in which NP[str]𝑗 follows NP[str]𝑘 . The first order corresponds to (36b) –repeated here as (81a) – and the second corresponds to (36c) – repeated here as(81b):


drengenboy.DEF

bliveris

givetgiven

boldenball.DEF

‘because the boy is given the ball’b. fordi

becauseboldenball.DEF

bliveris

givetgiven

drengenboy.DEF

‘because the ball is given to the boy’

Before turning to impersonal passives in Danish in the next subsection, I discussthe passive in double object constructions in Icelandic.

The distribution of structural/lexical case in Icelandic is basically the same asin German. The difference is that Icelandic allows for subjects with lexical caseand German does not. (82) shows our standard examples in Icelandic:


7 Passive

(82) ARG-ST

a. dansa (dance, unerg):⟨NP[str]

⟩b. lesa (read, trans):

⟨NP[str], NP[str]

⟩c. gefa (give, ditrans):

⟨NP[str], NP[ldat], NP[str]

⟩d. hjálpa (help, trans):

⟨NP[str], NP[ldat]

⟩The lexical rule in (79) licences the following participles:

(83)ARG-ST SPR COMPS

a. dansað (danced, unerg): ⟨⟩ ⟨⟩ ⟨⟩b. lesið (read, trans):

⟨NP[str]𝑗

⟩ ⟨NP[str]𝑗

⟩⟨⟩

c. gefið (given, ditrans):⟨NP[ldat]𝑗 , NP[str]𝑘

⟩ ⟨NP[ldat]𝑗

⟩ ⟨NP[str]𝑘

⟩⟨NP[str]𝑘 , NP[ldat]𝑗

⟩ ⟨NP[str]𝑘

⟩ ⟨NP[ldat]𝑗

⟩d. hjálpað (helped, trans):

⟨NP[ldat]𝑗

⟩ ⟨NP[ldat]𝑗

⟩⟨⟩

In addition to the ARG-ST list (83) shows the mapping to the SPR and COMPS fea-tures. Since Icelandic allows for quirky subjects the dative argument of ‘to help’can be mapped to the SPR list (Wechsler 1995: 147–148). Similarly the two ordersof the ARG-ST of ‘to give’ result in participles with a dative subject and a nomi-native subject as it is required for the analysis of (42a) and (42b) repeated hereas (84):

(84) a. Voruwere


gefnargiven

ambfittir?slaves.NOM

‘Was the king given slaves?’b. Var

wasambfittinthe.slave.NOM

gefingiven

konunginum?the.king.DAT

‘Was the slave given to the king?’

The impersonal passive with ‘to dance’ is parallel to the German impersonalpassive, but the passivization of ‘to help’ differs since this is an instance of thepersonal passive in Icelandic.

7.2.4.3 Impersonal passive

As a final point in this subsection let us have a look at the impersonal passive.German and Icelandic do not insist on subjects. So if there is no NP with struc-tural case, the construction in German is subjectless. Similarly Icelandic does not


7.2 The Analysis

require a subject: If there is no NP argument, the result is an impersonal passive.An example of the latter case is the passivization of dansa ‘to dance’. The ARG-STlist is the empty list and therefore the SPR list and the COMPS list are empty aswell. Passive participles of verbs that govern an NP and a PP object will have anARG-ST list that just contains the PP argument. This PP argument will be mappedto the COMPS list and hence a subjectless construction will result.

English does not allow for impersonal passives since it requires an NP or asentential argument that can serve as a subject. Danish requires a subject aswell, but allows for impersonal constructions. The trick that Danish employs isthe insertion of an expletive. I assume that the expletive insertion happens duringthe mapping of the ARG-ST elements to SPR and COMPS. If there is an NP/VP/CPat the beginning of the ARG-ST list, it is mapped to SPR and all other elements aremapped to COMPS. If there is no element that can be mapped to SPR, an expletiveis inserted.

(85) shows the mappings for Danish.

(85) ARG-ST SPR COMPS

a. danset/-s (unerg): ⟨⟩⟨NP𝑒𝑥𝑝𝑙

⟩⟨⟩

b. læst/-s (trans):⟨NP[str]𝑗

⟩ ⟨NP[str]𝑗

⟩⟨⟩

c. givet/-s (ditrans):⟨NP[str]𝑗 , NP[str]𝑘

⟩ ⟨NP[str]𝑗

⟩ ⟨NP[str]𝑘

⟩⟨NP[str]𝑘 , NP[str]𝑗

⟩ ⟨NP[str]𝑘

⟩ ⟨NP[str]𝑗

⟩d. hjulpet/-s (trans):

⟨NP[str]𝑗

⟩ ⟨NP[str]𝑗

⟩⟨⟩

7.2.4.4 The passive auxiliary

We now saw how the participle forms of the languages under considerationslook like and how they are licensed from lexical entries for stems via lexicalrules. What is missing is the lexical items for the auxiliary verbs. Chapter 5 dealtwith the analysis of verbal complexes in SOV languages like German, Dutch andAfrikaans and it was pointed out that SVO languages like English and the Scan-dinavian languages do not form a verbal complex. With this background it maycome as a surprise that it is possible to formulate one general constraint cover-ing all the passive auxiliaries in the Germanic languages. The following AVMshows a constraint holding on all ARG-ST lists for passive auxiliary in Germaniclanguages:8

8The lexical item of the passive auxiliary used byMüller (2002: 147) andMüller & Ørsnes (2013a:149) specifies the DA value of the embedded participle to be a referential NP. This excludes the


7 Passive

(86)Passive auxiliary for Germanic languages (be, werden, blieve, etc.):ARG-ST 1 ⊕ 2 ⊕

⟨VFORM pppSPR 1COMPS 2

⟩

When the passive auxiliary in (86) is used in a German grammar, the argu-ments of the participle ( 1 and 2 ) are attracted by the passive auxiliary (Hinrichs& Nakazawa 1989b; 1994). Auxiliaries in the SVO langauges embed a VP. Thismeans that the COMPS value of the embedded verb has to be the empty list. There-fore 2 is the empty list and only the specifier ( 1 ) is taken over from the embeddedverb.

With the lexical item for the auxiliary in place, we can have a look at someexample analyses of passive sentences. Let’s start with an English example. Theanalysis of (87) is shown in Figure 7.3.

(87) The child was given a novel.

The lexical item for the stem give- is input to the passive lexical rule. Thepassive lexical rule licenses the participle given. The ARG-ST of give- is shortenedby the element in the DA list of give- ( 4 ). The result is an ARG-ST list containingthe two NPs that would be the objects in active sentences. The first one ( 1 ) ismapped to SPR and the second one ( 3 ) to COMPS. The combination of given anda novel forms a VP (something with an empty COMPS list and an element in theSPR list). The passive auxiliary was selects the VP given a novel. The specifierof the selected VP ( 1 ) is attracted. As the first argument of was with structuralcase, this NP gets nominative. Finally, the VP was given a novel is combined withthe child and we have a complete sentence.

The analysis of the parallel German sentence in (88) is shown in Figure 7.4 onpage 162.

(88) dassthat

demthe

Kindchild

eina

Romannovel

gegebengiven

wurdewas

(German)

‘that the child was given a novel’

The analysis is similar to the one of the English example but it differs in that theauxiliary and the participle is forming a verbal complex. First, the passive lexicalrule applies to a verb stem geb- and licenses the particple form gegeben. gegeben( 3 ) is combined with wurde and wurde takes over the elements of the COMPS listof gegeben ⟨ 1 , 2 ⟩. The result of the combination of gegeben and werden has theCOMPS list ⟨ 1 , 2 ⟩.

passivization of unaccusative verbs, which have the empty list as the DA value.


7.2 The Analysis

V[SPR ⟨ ⟩,COMPS ⟨ ⟩]

1 NP[nom]

the child

V[SPR ⟨ 1 ⟩,COMPS ⟨ ⟩]

V[SPR ⟨ 1 ⟩,COMPS ⟨ 2 ⟩,ARG-ST ⟨ 1 , 2 ⟩]

was

2 V[SPR ⟨ 1 ⟩,COMPS ⟨ ⟩]

V[SPR ⟨ 1 ⟩,COMPS ⟨ 3 ⟩,ARG-ST ⟨ 1 , 3 ⟩]

V[DA ⟨ 4 ⟩,ARG-ST ⟨ 4 , 1 , 3 ⟩]

give-

3 NP[acc]

a novel

Figure 7.3: Analysis of The child was given a novel.

7.2.4.5 The morphological passive

A lexical rule similar to the one accounting for the participle forms can be usedfor the morphological passives in Danish. One difference is of course the affixalmaterial that is added by the rule. Furthermore, it is assumed for the morpholog-ical passive that the DA of the input to the lexical rule has to contain a referentialXP. As was discussed in the previous section, this excludes morphological pas-sives of unaccusatives and weather verbs.

7.2.4.6 Perfect

Haider’s (1986a) analysis of the passive is brilliant since it is sufficient to have onelexical item for the participle. The participle has a blocked designated argumentand in the passive the designated argument remains blocked, while the prefectauxiliary deblocks the designated argument.


7 Passive

V[SPR ⟨ ⟩,COMPS ⟨ ⟩]

1 NP[dat]

dem Kindthe child

V[SPR ⟨ ⟩,COMPS ⟨ 1 ⟩]

2 NP[nom]

ein Romana novel

V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩]

3 V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 ⟩,ARG-ST ⟨ 1 , 2 ⟩]

V[DA ⟨ 4 ⟩,ARG-ST ⟨ 4 , 1 , 2 ⟩]

geb-give-

V[SPR ⟨ ⟩,COMPS ⟨ 1 , 2 , 3 ⟩,ARG-ST ⟨ 1 , 2 , 3 ⟩]

wurdewas

Figure 7.4: Analysis of dass dem Kind ein Roman gegeben wurde ‘that the childwas given a novel’

(89) a. dassthat

derthe.NOM

Aufsatzpaper

gelesenread

wurdewas

‘that the paper was read’b. dass

thatKirbyKirby

denthe.ACC

Aufsatzpaper

gelesenread

hathas

‘that Kirby has read the paper’

Deblocking of the designated argument is possible since the designated argu-ment is encoded not just in the stem of a verb but also in the lexical item for theparticiple.

(90) Lexical item for perfect auxiliaries in SOV languages like Dutch andGerman:


7.2 The Analysis

ARG-ST 1 ⊕ 2 ⊕ 3 ⊕⟨

VFORM pppDA 1SPR 2COMPS 3

⟩

Unfortunately Haider’s approach does not work for SVO langauges. If wewanted to use the argument blocking/deblocking approach, we had to assumethe structures in (91a–b):

(91) a. He [has given] the book to Mary.b. The book [was given] to Mary.c. He has [given the book to Mary].d. The book was [given to Mary].

If we assume that auxiliaries embed VPs as was argued on page 102, we run intoproblems since the subject of the participle is blocked and the only VP we canform with the participle is the one in (91d), but for the perfect we also need a VPcontaining the object as in (91c).

7.2.4.7 The remote passive

The so-called remote passive is a highlight of German syntax since several phe-nomena interact in a non-trivial way. It was first discussed by Höhle (1978: 175–176). Höhle observed that objects of German infinitives with zu appear in thenominative in certain contexts. (92) provides some constructed examples fromthe literature:

(92) a. daßthat

erhe.NOM

auchalso

vonfrom

mirme

zuto

überredenpersuade

versuchttried

wurde9

got(German)

‘that an attempt to persuade him was also made by me’b. weil

becausederthe.NOM

Wagencar

oftoften

zuto

reparierenrepair

versuchttried

wurdewas

‘because many attempts were made to repair the car’

The examples in (93) are attested data collected from the COSMAS corpus byMüller (2002: 136–137):

(93) a. Dabei darf jedoch nicht vergessen werden, daß in derBundesrepublik, wo ein Mittelweg zu gehen versucht wird, die

9Oppenrieder (1991: 212).


7 Passive

Situation der Neuen Musik allgemein und die Stellung derKomponistinnen im besonderen noch recht unbefriedigend ist.10

‘One should not forget that the situation of the New Music in generaland the position of female composers in particular is ratherunsatisfying in the Bundesrepublik, where one tries to follow amiddle course.’

b. Noch ist es nicht so lange her, da ertönten gerade aus dem Thurgaujeweils die lautesten Töne, wenn im Wallis oder am Genfersee imUmfeld einer Schuldenpolitik mit den unglaublichsten Tricks dersportliche Abstieg zu verhindern versucht wurde.11

‘It still is not too long ago that the loudest protests were heard in theThurgau itself when the most unbelievable tricks in the sphere ofdebt policies were applied to prevent relegation in the Valais or atLake Geneva.’

c. Die Auf- und Absteigenden erzeugen ungewollt einen Ton, derbewusst nicht als lästig zu eliminieren versucht wird, sondern zumEigenklang des Hauses gehören soll, so wünschen es sich dieArchitekten.12

‘The people who go up and down produce a tune without intentionwhich is not consciously sought to be eliminated but which, rather,belongs to the individual sound of the building, as the architectsintended.’

Höhle’s examples, other examples from the literature involved the verb versuchen‘to try’, butWurmbrand (2003b) showed that other verbs allow for the remote pas-sive as well. (94) shows some of her examples with beginnen ‘to start’, vergessen‘to forget’, and wagen ‘to dare’:

(94) a. derthe.NOM

zweitesecond

Entwurfplan

wurdewas

zuto

bauenbuild

begonnen,13

started‘It was begun to build the second plan.’

While the case of der zweite Entwurf ‘the second plan’ is unambiguously nom-inative, this is not the case for the examples in (95), since the respective elements

10Mannheimer Morgen, 26.09.1989, Feuilleton; Ist’s gut, so unter sich zu bleiben?11St. Galler Tagblatt, 09.02.1999, Ressort: TB-RSP; HCT und das Prinzip Hoffnung.12Züricher Tagesanzeiger, 01.11.1997, p. 61.13http://www.waclawek.com/projekte/john/johnlang.html, 28.07.2003.


http://www.waclawek.com/projekte/john/johnlang.html

7.2 The Analysis

are in the plural and hence could be nominative or accusative. But due to agree-ment with the finite verb, it is clear that the relative pronoun are in the nomina-tive.

(95) a. Anordnungen, die zu stornieren vergessen wurden14

b. Aufträge […], die zu drucken vergessen worden sind15

c. NUR Leere, oder doch noch Hoffnung, weil aus Nichts wiederGefühle entstehen, die so vorher nicht mal zu träumen gewagtwurden?16

d. Dem Voodoozauber einer Verwünschung oder die gefaßteEntscheidung zu einer Trennung, die bis dato noch nichtauszusprechen gewagt wurden17

The object of a verb that is embedded under a passive participle is promotedto subject of the sentence:

(96) a. weilbecause

AickeAicke

denthe.ACC

Wagencar

oftoften

zuto

reparierenrepair

versuchttried

hathas(German)

b. weilbecause

derthe.NOM

Wagencar

oftoften

zuto

reparierenrepair

versuchttried

wurdewas

‘because many attempts were made to repair the car’

The remote passive is possible in verbal complexes only. If no verbal complexis formed as in (97a,c), the object of reparieren has to appear in the accusative:

(97) a. weilbecause

oftoften

versuchttried

wurde,was

denthe.ACC

Wagencar

zuto

reparieren.repair

‘because many attempts were made to repair the car.’b. * weil

becauseoftoften

versuchttried

wurde,was

derthe.NOM

Wagencar

zuto

reparieren.repair

c. Denthe.ACC

Wagencar

zuto

reparierenrepair

wurdewas

oftoften

versucht.tried

d. * Derthe.NOM

Wagencar

zuto

reparierenrepair

wurdewas

oftoften

versucht.tried

14http://www.rlp-irma.de/Dateien/Jahresabschluss2002.pdf, 28.07.2003.15http://www.iitslips.de/news.html, 28.07.2003.16http://www.ultimaquest.de/weisheiten_kapitel1.htm, 28.07.2003.17http://www.wedding-no9.de/adventskalender/advent23_shawn_colvin.html, 28.07.2003.


http://www.rlp-irma.de/Dateien/Jahresabschluss2002.pdf

http://www.iitslips.de/news.html

http://www.ultimaquest.de/weisheiten_kapitel1.htm

http://www.wedding-no9.de/adventskalender/advent23_shawn_colvin.html

7 Passive

The difference between (96b) and (97a,c) are explained by an analysis that treatsthe remote passive as a passivization of a predicate complex, i.e., by an analysisthat assigns the structure (98) to (96b).

(98) weilbecause

derthe.NOM

Wagencar

oftoften

[[zuto

reparierenrepair

versucht]tried

wurde].was

‘because many attempts were made to repair the car.’

In (97a,c) we do not have predicate complexes. The object of zu reparieren is partof the VP and therefore it gets accusative. The passives in (97a,c) are impersonalpassives.

The verb versuchen ‘to try’ selects a subject, an infinitive with zu ‘to’ and thecomplements of the embedded verb.

(99) versuch- ‘to try’:[ARG-ST

⟨NP[str]𝑖

⟩⊕ 1 ⊕

⟨V[inf, SUBJ

⟨NP[str]𝑖

⟩, COMPS 1

]⟩]In our example, the embedded verb is reparieren and has one complement. (100)

shows the ARG-ST value of versuch-. The first NP is the subject of versuchen andthe second NP is the attracted object of zu reparieren.

(100) ARG-ST value of versuch- with embedding of a strictly transitive verb:⟨NP[str]𝑖 , NP[str]𝑗 , V[inf ]

⟩When the passive lexical rule applies to this verb stem and licenses the respectiveparticiple, the resulting lexical item for versucht ‘tried’ will have the followingARG-ST value:

(101) ARG-ST of participle versucht with embedding of strictly transitive verb:⟨NP[str]𝑗 , V[inf ]

⟩The first NP is mapped to SUBJ and V[inf ] is mapped to COMPS. After combina-tion with zu reparieren, we have a complex with NP[str]𝑗 in SUBJ and nothing inCOMPS. wurde selects a verb or verbal complex with the verb form ppp and thearguments of the embedded verb. Hence, NP[str]𝑗 ends up as the first elementof the ARG-ST list of werden where it gets nominative. Since werden is finite allARG-ST elements are mapped to COMPS and have to be realized in the sentence.The analysis of the verbal complex in (98) is shown in Figure 7.5 on the next page.

The remote passive is also possible with object control verbs, that is, verbstaking a subject and an object and a verbal projection the subject of which iscoreferential with the object. An example is erlauben ‘to permit’. (102a) shows theverb in the active and without verbal complex formation. The object of erlauben


7.2 The Analysis

V[VFORM fin,COMPS 1 ]

2 V[VFORM ppp,SUBJ 1 ,COMPS ⟨⟩]

3 V[VFORM inf ,SUBJ

⟨NP[str]𝑖

⟩,

COMPS 1

⟨NP[str]𝑗

⟩]

zu reparierento repair

V[VFORM ppp,SUBJ 1 ,COMPS ⟨ 3 ⟩ ]

versuchttried

V[VFORM fin,COMPS 1 ⊕ ⟨ 2 ⟩]

wurdewas

Figure 7.5: The analysis of the remote passive as passivization of a complex form-ing verb

‘to permit’ uns ‘us’ is coreferential with the subject of den Erfolg auszukosten ‘toenjoy the success’. (102b,c) show that the object of auszukosten can be realizedin the nominative, if the verbs are forming a verbal complex:

(102) a. Siethey

erlaubenpermitted

unsus.DAT

nicht,not

denthe.ACC

Erfolgsuccess

auszukosten.to.enjoy

‘They did not permit us to enjoy the success.’b. Keine

noZeitungnewspaper.NOM

wirdwas

ihrher.DAT

zuto

lesenread

erlaubt.18

allowed‘She is not allowed to read any newspapers.’

c. Derthe

Erfolgsuccess.NOM

wurdewas

unsus.DAT

nichtnot

auszukostento.enjoy

erlaubt.19

permitted‘We were not permitted to enjoy our success.’

The passive of the construction without verbal complex is an impersonal passive:

(103) Unsus.DAT

wurdewas

erlaubt,allowed

denthe.ACC

Erfolgsuccess


18Stefan Zweig. Marie Antoinette. Leipzig: Insel-Verlag. 1932, p. 515, quoted from Bech (1955: 309).That this is an instance of the remote passive was noted by Askedal (1988: 13).

19Haider (1986b: 110).


7 Passive

(104) shows the ARG-ST value of erlaub-: erlauben takes a subject and a dativeobject. The dativ object is coindexed with the subject of the embedded verb, thatis, the two NPs have the same index, namely 𝑗 .

(104) erlaub- ‘to permit’:[ARG-ST

⟨NP[str]𝑖 , NP[ldat]𝑗

⟩⊕ 1 ⊕

⟨V[inf, SUBJ

⟨NP[str]𝑗

⟩, COMPS 1

]⟩]The complements of the embedded verb ( 1 ) are taken over by the embeddingverb. (105) shows the ARG-ST value of the respective highest verb:

(105) a. zu lesen erlauben:⟨NP[str]𝑖 , NP[ldat]𝑗 , NP[str]𝑘 , V[COMPS

⟨NP[str]𝑘

⟩]⟩

b. zu lesen erlaubt wird:⟨NP[ldat]𝑗 , NP[str]𝑘 , V[COMPS

⟨NP[str]𝑘

⟩]⟩

(105a) shows how the object of lesen is attracted so that the combined ARG-STcontains three NPs. (105b) shows the passive variant in which the subject oferlauben is suppressed. The result is a ARG-ST list starting with a dative NP, anNP with lexical case. Since the first NP with structural case gets nominativeand agrees with the finite verb, the theory makes the right predictions even insituations as complex as the remote passive with object control verbs. The firstNP with structural case is the subject in German.

7.3 Summary

Summing up, it can be said that this chapter provides a unified account of thepassive in Danish, Englisch, German, and Icelandic. The lexical rule accountsboth for the morphological and the analytical passives. The first element on theARG-ST list is suppressed and a relational constraint promote promotes any NPwith structural case. The languages differ in what cases they use and which casesare structural/lexical. Danish inserts expletives to allow for impersonal passivesand fulfilling the need of a subject. This expletive insertion is done in the ARG-STmapping when arguments are mapped from ARG-ST to the valence lists.

The SVO languages seem to require different items for the perfect/passive par-ticiples, but Haider’s (1986a) passive analysis for German using just one participleform for bot perfect and passive can be maintained.


7.3 Summary


1. What tests do you know for subjecthood?

2. Do these tests work for all Germanic languages?

3. In which way is German different from Icelandic in terms of subjects?

4. What is structural case? What is lexical case?

5. What is an impersonal passive?

6. Does Icelandic have impersonal passives?


7 Passive

Exercises

1. Which NPs in (106) do have structural and which lexical case?

(106) a. Derthe.NOM

Jungeboy

lacht.laughs

b. MichI.ACC

friert.freeze

‘I am cold.’c. Er

he.nomzerstörtdestroys

dasthe.ACC

Auto.car

‘He destroys the car.’d. Das

this.NOMdauerttakes

eina.acc

ganzeswhole

Jahr.year

‘This takes a whole year.’e. Er

he.NOMhathas

nurjust

einenone.ACC

Tagday

dafürthere.for

gebraucht.needed

‘He needed a day for this.’


7.3 Summary

f. Erhe.NOM

denktthinks

anat

denthe.ACC

morgigentomorrow

Tag.day

‘He thinks about tomorrow.’

2. Give ARG-ST lists for the following verbs:

(107) a. show, read, meet (English)b. zeigen, begegnen, treffen (German)c. (Danish)d. (Icelandic)

If you are uncertain as far as case is concerned, you may use the Wik-tionarya.

3. Draw the analysis tree for the following sentence:

(108) that the box was opened

Please provide valence features (SPR and COMPS) and part of speech in-formation. You may abbreviate the NP using a triangle.

ahttps://de.wiktionary.org/, 2018-07-02.



8 Clause types and expletives

Chapter 6 discussed the analysis of verb-initial and verb-second sentences. Thischapter deals with embedded sentences with complementizers (e.g., that in En-glish) and embedded interrogative sentences.

8.1 The phenomena

The phenomenon section consists of three parts: Section 8.1.1 deals with embed-ded clauses introduced by a complementizer, Section 8.1.2 describes the structureof interrogative clauses and Section 8.1.3 deals with expletives in V2 and inter-rogative clauses.


As was already mentioned Afrikaans, Dutch, German are SOV languages andthis is also shown in embedded clauses that are introduced by a complementizer.(1) is an example:

(1) IchI

weiß,know

dassthat

AickeAicke

dasthe

Buchbook

heutetoday

gelesenread

hat.has

(German)

‘I know that Aicke read the book today.’

English, beeing an SVO non-V2 language, allows for SVO order only.

(2) I know that Kim has read the book yesterday. (English)

Interestingly, Danish, also an SVO languauge, allows both SVO order (3) andV2 order (4) in clauses preceeded by a complementizer:

(3) JegI

ved,know

atthat

GertGert

ikkenot

harhas

læstread

bogenbook.DEF

i dag.today

(Danish)

‘I know that Gert did not read the book today.’

(4) a. JegI

ved,know

atthat

i dagtoday

harhas

GertGert

ikkenot

læstread

bogen.book.DEF

(Danish)


b. JegI

ved,know

atthat

bogenbook.DEF

harhas

GertGert

ikkenot

læstread

i dag.today

The example in (3) includes the negation in order to show that we indeed dealwith the SVO order here. Without the negation it is not clear whether non-V2clauses are allowed in clauses that are introduced by a complementizer since (5a)has the finite verb in second position. With the negation present, it is clear thatwe have a V2 clause if the negation follows the finite verb as in (5b) and that wedo not have a V2 clause if the finite verb follows the negation as in (3) and henceis in third position.

(5) a. atthat

GertGert

harhas

læstread

bogenbook.DEF

(V2 or SVO)

b. atthat

GertGert

harhas

ikkenot

læstread

bogenbook.DEF

(V2)

For complementizerless sentences the V2 order is the only one that is possible:

(6) a. GertGert

harhas

ikkenot

læstread

bogenbook.DEF

(V2)

b. * GertGert

ikkenot

harhas

læstread

bogenbook.DEF

(SVO)

Yiddish and Icelandic are SVO languages as well. The clauses that are com-bined with a complementizer are V2:

(7) a. IkhI

meynthink

azthat

haynttoday

hothas

MaxMax

geleyentread

dosthe

bukh.1

book(Yiddish)

‘I think that Max read the book today.’b. Ikh

Imeynthink

azthat

dosthe

bukhbook

hothas

MaxMax

geleyent.read

(8) Engumno.one.DAT

dattfell

íto

hug,mind

aðthat

vertworth

væriwas

aðto

reynatry

tilPREP

aðto

kynnastknow

honum.2

him

(Icelandic)

‘It didn’t occur to anyone that it was worth trying to get to know him.’

1Diesing (1990: 58)2Maling (1990: 75)


8.1 The phenomena


TheOV languages form subordinated interrogative clauses by preposing a phrasecontaining an interrogative pronoun3 from an otherwise SOV clause. (9) showsa German example:

(9) a. IchI

weiß,know

werwho

heutetoday

dasthe

Buchbook

gelesenread

hat.has

(German)

‘I know who read the book today.’b. Ich

Iweiß,know

waswhat

AickeAicke

heutetoday

gelesenread

hat.has

‘I know what Aicke has read today.’

Since languages like German allow for scrambling, senteces like those in (9) couldjust be due to the permutation of arguments of a head. However, the generaliza-tion about these w-clauses is that an arbitrary w-element can be fronted. (10)gives a German example involving a nonlocal dependency:

(10) IchI

weißknow

nicht,not

[überabout

welcheswhich

Thema]𝑖topic

sieshe

versprochenpromised

hat,has

[[einena

Vortragtalk

_𝑖]to

zuhold

halten].

(German)

‘I do not know about which topic she promised to give a talk.’

Here, the phrase über welches Thema ‘about which topic’ is an argument of Vor-trag, which is embedded in the VP containing zu halten ‘to hold’, which is in turnembedded under versprochen hat ‘promised has’. The generalization about inter-rogative clauses is that an interrogative clause consists of a interrogative phrase(über welches Thema ‘about which topic’) and a clause in which this interroga-tive phrase is missing somewhere (sie versprochen hat, einen Vortrag zu halten‘she promised to give a talk’).

In German the order of the other constituents is free as in assertive mainclauses and embedded clauses with a complementizer that were discussed ear-lier.

(11) a. IchI

weiß,know

waswhat

keinernobody

diesemthis


gebengive

würde.would

(German)

‘I know what nobody would give this man.’

3Most interrogative pronouns start withw in German andwh in English. Phrases containing aninterrogative pronoun are called w-phrases or wh-phrases, respectively. Interrogative clausesare sometimes called w-clauses or wh-clauses.



b. IchI

weiß,know

waswhat

diesemthis


keinernobody

gebengive

würde.would

This follows from that we said so far, since interrogatives are just SOV sentenceswith one constituent extracted. The possibility of scrambling constituents is notaffected by extracting a phrase.

In Danish and English the interrogative clauses consist of an interrogativephrase and an SVO clause in which it is missing:

(12) a. GertGert

harhas

givetgiven

hamhim

bogen.book.DEF

(Danish)

‘Gert gave him the book.’b. Jeg

Ived,know

hvad𝑖what

[GertGert

harhas

givetgiven

hamhim

_𝑖].

‘I know what Gert gave him.’c. Jeg

Ived,know

hvem𝑖

who[GertGert

harhas

givetgiven

_𝑖 bogen].book.DEF

‘I know who Gert has given the book.’

(12a) shows the clause with SVO order and (12b) is an example with the secondaryobject as interrogative ponoun and (12c) is an example with the primary objectas interrogative pronoun. The position that the respective objects have in non-interrogative clauses like (12a) is marked with _𝑖 .

Yiddish is special in that it has a V2 order in interrogative clauses as well(Diesing 1990: Sections 4.1, 4.2): interrogatives consist of a interrogative phrasethat is extracted from a V2 clause:

(13) Iryou

veystknow

efshermaybe

[avuwhere

dothere

voyntlives

RoznblatRoznblat

derthe

goldshmid]?4

goldsmith‘Do you perhaps know where Roznblat the goldsmith lives?’

So the variation is w-phrase + SOV, w-phrase + SVO, and w-phrase + V2.In addition to the question regarding the order within the embedded clause

(SOV, SVO, V2), there is variation in finiteness of the embedded clause:

(14) a. I wonder what to read.b. I wonder what I should read.c. * Ich

Ifrageask

mich,myself

waswhat

zuto

lesen.read

(German)

4Diesing (1990: 65). Quoted from Olsvanger, Royte Pomerantsn, 1949.


8.1 The phenomena

d. IchI

frageask

mich,myself

waswhat

ichI

lesenread

sollshall

/ kann.can

(German)

‘I wonder what I shall/can read.’ or ‘I wonder what to read.’

English allows for infinitiveswith to. In comparison to finite interrogative clausesthe infinitival form adds a modal meaning. German does not allow for non-finiteinterrogatives as (14c) shows.

8.1.3 The use of expletives to mark the clause type

The Germanic languages use constituent order to code the clause type: V2 mainclauses can be assertions or questions, depending on the content of the preverbalmaterial and intonation. Similarly embedded interrogative clauses consist of aw-phrase and an SVO, SOV, or V2 clause. The fronting of a constituent in a V2clause comes with certain information structural effects: something is the topicor the focus of an utterance. For embedded sentences it is important for somelanguages that the structure is transparent that is, that we have thew + SVO orw+ V2 order. There are situations in which it is inappropriate to front an elementand in such situations the Germanic languages use expletives, that is, pronounsthat do not contribute semantically, to maintain a certain order.

German uses the expletive es to fill the position in front of the finite verb, ifno other constituent is to be fronted.

(15) a. Dreithree

Reiterriders

rittenrode

zumto.the

Torgate

hinaus.out

(German)

‘Three riders rode out of the gate.’b. Es

EXPLrittenrode

dreithree

Reiterriders

zumto.the

Torgate

hinaus.out

Danish uses the expletive to make it clear that an extraction of a consituenttook place (Müller & Ørsnes 2011: 169):5

(16) a. Politietpolice.DEF

vedknows

ikke,not

hvemwho

derEXPL

havdehas

placeretplaced

bomben.6

bomb.DEF

(Danish)

‘The police does not know who placed the bomb.’5Examples marked with DK are extracted from KorpusDK, a corpus of 56 Million words docu-menting contemporary Danish (http://ordnet.dk/korpusdk).

6DK


http://ordnet.dk/korpusdk


b. * Politietpolice.DEF

vedknows

ikke,not

hvemwho

havdehas

placeretplaced

bomben.bomb.DEF

Without the expletive one would have a pattern like the one in (16b). In (16b) wehave the normal SVO order and it is not obvious to the hearer that the patternconsists of an extracted element (the subject) and an SVO clause from which itis missing. This is more transparent if an expletive is inserted into the subjectposition as in (16a). (17) shows this using the analysis that will be suggestedin Section 8.2.3: (17a) shows the hypothetical structure that would result if oneassumed that the subject hvem ‘who’ is extracted. So-called string-vacuous move-ment would result: the subject is moved to a place right next to it. In (17b) on theother hand the subject position is taken by the expletive and hence it is clear thatthe embedded sentence has a special structure. There is an overt marker for thehearer or reader of the sentence marking it as an embedded interrogative clause.

(17) a. * [hvem𝑖

who[_𝑖 havde

hasplaceretplaced

bomben]]bomb.DEF

(Danish)

b. [hvem𝑖

who[derEXPL

havdehas

_𝑖 placeretplaced

bomben]]bomb.DEF

Similarly Yiddish uses an expletive in embedded interrogatives (w + V2) ifthere is no other element that is information structurally appropriate for thepreverbal position. (18) shows examples from Prince (1989: 403–404):

(18) a. ikhI

hobhave

ziher

gefregtasked

verwho

esEXPL

izis

beserbetter

farfor

irher

(Yiddish)

‘I have asked her who is better for her.’b. ikh

Ihobhave

imhim

gefregtasked

vemenwhom

esEXPL

kenenknow

aleall

dayneyour

khaverimfriends

‘I asked him whom all your friends know.’

(18a) is an example involving an interrogative pronoun that is the subject and(18b) is an example in which the preverbal position is not filled by an argumentof kenen ‘know’ but by an expletive. The subject ale dayne khaverim ‘all yourfriends’ stays behind and the object vemen ‘whom’ is extracted since it is theinterrogative pronoun.


8.2 The analysis

8.2 The analysis


The analysis of Afrikaans, Dutch, German and English complementizer phrasesis straightforward: the complementizer is combined with an uninverted verbalprojection. For the first languages this is a verb last clause (SOV) and for Englishit is an SVO clause. The respective analyses are given in Figure 8.1 and 8.2.

CP

C

dassthat

S

NP[nom]

niemandnobody

V′

NP[acc]

ihnhim

V

kenntknows

Figure 8.1: Analysis of German complementizer phrase as C + VL

CP

C

that

S

NP[nom]

nobody

VP

V

knows

NP[acc]

him

Figure 8.2: Analysis of English complementizer phrase as C + SVO

Yiddish complementizers select for a V2 clause. The analysis of the example



in (19) is shown in Figure 8.3.

(19) IkhI

meynthink

azthat

haynttoday

hothas

MaxMax

geleyentread

dosthe

bukh.7

book(Yiddish)

‘I think that Max read the book today.’

CP

C

azdass

S

Adv𝑖

haynttoday

S/Adv

V ⟨ S//V ⟩

V

hot𝑘has

S//V/Adv

NP

MaxMax

VP//V/Adv

V//V

_𝑘

VP/Adv

VP

V

geleyentread

NP

dos bukhthe book

Adv/Adv

_𝑖

Figure 8.3: Analysis of the Yiddish complementizer phrase as C + V2

The analysis looks complicated but it is really just the combination of a comple-mentizer with a V2 clause. The V2 clause has haynt ‘today’ extracted and thefinite auxiliary hot ‘has’ is moved to V1 position.

The differences between languages can be accounted for by letting the com-plementizer select Ses with different feature–value combinations. While comple-mentizers in SOV languages select for verb-final projections (INITIAL−), Yiddishselects for V2+ sentences (the result of applying the Filler-Head Schema) and

7Diesing (1990: 58)


8.2 The analysis

Danish does not specify any of such features on the selected clause. Since noth-ing is specified in Danish, the embedded clause can have the form that the restof the Danish grammar permits: it can be SVO or V2.


As the data discussion showed, the phrase containing the interrogative pronounis extracted from the remaining clause. The fronting of the w-phrase is likefronting in V2 sentences. In the topological fields model, the fronted phrasein German relative clauses and interrogative clauses is assigned to the Vorfeld(Müller 2020: 47–49). The difference between interrogative clauses andV2 clausesis the position of the verb: V2 sentences have the verb in initial position while itis in final position in interrogatives and relatives. Figure 8.4 shows the analysisof the interrogative clause in (20). The pronominal adverb worüber ‘about what’is extracted from the rest of the clause.

(20) IchI

weiß,know

[worüber]𝑖where.about

[ _𝑖 sieshe

spricht].speaks

‘I know what she speaks about.’

S

PP

worüberwhere.about

S/PP

PP/PP

_

V′

NP

sieshe

V

sprichtspeaks

Figure 8.4: Analysis of simple interrogative clause

Figure 8.4 uses the slash notation that I have been using so far. In order to ac-count for more complex w-phrases I will use the same trick as for other nonlocaldependencies and pass information about w-pronouns on to mother nodes. Formodeling this I will use list-valued features like SPR and COMPS. Figure 8.5 shows



the same sentence as Figure 8.4 but with two features that are traditionally usedin the analysis of nonlocal dependencies (Pollard & Sag 1994: Chapter 4 and 5):qUE and SLASH. By convention, the boxed numbers are put in front of XPs if theXP is an argument and they follow the XP if the XP is involved in a nonlocaldependency. The reason for this is that different parts of information are shared.A full explanation of the difference requires some deeper understanding of themechanisms and cannot be given here. The interested reader is referred toMüller(2013b: Chapter 10) or to Borsley & Crysmann (2021). PP 1 [SLASH ⟨ 1 ⟩] meansthat the relevant information about the PP is put into SLASH, that is, into the listthat is percolated upwards until a matching filler is found (a PP whose relevantproperties can be identified with the element in the SLASH list). Once a filler hasbeen found no SLASH element is passed upwards. The SLASH list of the top-mostnode is the empty list.

S[SLASH ⟨ ⟩]

PP 1

worüberwhere.about

S[SLASH ⟨ 1 ⟩]

PP 1 [SLASH ⟨ 1 ⟩]

_

V′

NP

sieshe

V

sprichtspeaks

Figure 8.5: Analysis of simple interrogative clause using the SLASH feature

Now, this machinery can be extended to cover nonlocal dependencies for inter-rogative pronouns. Figure 8.6 shows how the information about the interrogativewithin the complex w-phrase can be passed upwards in a tree by using the qUEfeature.

(21) IchI

weiß,know

[überabout

welcheswhich

Thema]𝑖topic

[ _𝑖 sieshe

spricht].speaks

‘I know which topic she speaks about.’

Figure 8.6 is completely parallel to Figure 8.4 except that information about thew-word is added. The content of qUE is not provided here but the qUE list of w-


8.2 The analysis

words contains information that is needed for semantics: the w-word indicateswhat is asked for and this information is passed up to the level of the completeclause (see Ginzburg & Sag 2000 on interrogatives in general and on their seman-tics in particular).

S[qUE ⟨ ⟩, SLASH ⟨ ⟩]

PP 1 [qUE ⟨ 2 ⟩]

P

über

NP[qUE ⟨ 2 ⟩]

Det[qUE ⟨ 2 ⟩]

welches

N

Thema

S[SLASH ⟨ 1 ⟩]

PP 1 [SLASH ⟨ 1 ⟩]

_

V′

NP

sie

V

spricht

Figure 8.6: Analysis of simple interrogative clause using SLASH and qUE

Interrogative clauses are licensed by a special variant of the Filler-Head Schema,namely a schema that requires the initial daughter (the filler) to have somethingin its qUE list (Figure 8.7). This entails that the filler has to contain a w-word.

H[qUE ⟨ ⟩, SLASH ⟨ ⟩]

1 [qUE ⟨ ⟩] H[SLASH ⟨ 1 ⟩]

Figure 8.7: Interrogative Clause Schema

The languages differ as far as the order of the verb and its arguments are con-cerned, so further specifications have to be added to what is given in Figure 8.7.For example, German interrogative clauses are verb final, while Yiddish interrog-atives involve an extraction out of a V2 clause. In addition, constraints regardingthe verb form have to be specified. German allows for finite verbs only, whileEnglish allows for finite verbs and infinitives with to (see (14) on p. 176).

Up to now, we have looked at German examples with prepositional objectsfronted. Figure 8.8 shows the analysis of (22) with a subject as a w-phrase:



(22) werwho

dasthe

Buchbooks

liestreads

(German)

‘who reads the book’

S

NP[snom]

werwho

S/NP

NP/NP

_

V′

NP[sacc]

das Buchthe book

V

liestreads

Figure 8.8: Analysis of German interrogative clause with the subject as w-word

So, with what we have so far, we can analyze interrogatives in German andother SOV languages but there are still open questions in languages like Dan-ish where expletive insertion in subject position is required when a subject isquestioned (see (16)). Similarly, Yiddish may insert expletives in the preverbalposition in interrogatives (see (18)). So, to account for this data, we have to dealwith expletives. Expletives are the topic of the next section.

8.2.3 A lexical rule for the introduction of expletives

The various types of expletives introduced in Section 8.1.3 can – maybe some-what surprisingly – be accounted for by a simple lexical rule that adds an exple-tive to the ARG-ST list of lexical items (Müller & Ørsnes 2011: 180):

(23) Expletive Insertion Lexical Rule:HEAD

[verbVFORM fin

]ARG-ST 1

↦→[ARG-ST

⟨NP[lnom]expl

⟩⊕ 1

]The application of the lexical rule is restricted to finite verbs since positionalexpletives occur in V2 sentences and these are always finite. Expletives in inter-


8.2 The analysis

rogative clauses are used to fill the subject position to mark extraction and ofcourse this is something that is necessary in finite clauses only.

The case of the expletive pronoun is specified as lexical nominative, whichmeans that it is invisible to case assignment principles. Nominative is assignedto the first NP with structural case (see p. 148) and since the expletive has lexicalcase, nothing changes. The same is true for structural accusatives: the first NPwith structural gets nominative and all others accusative. The expletive does notinterfere with this.

Similarly, the theory of agreement entertained so far is not affected: the verbagrees with the first NP with structural case. This makes the right predictionsfor agreement in Icelandic, where the verb agrees with objects in the nominative(Zaenen et al. 1985: 460).

(24) a. Hefurhas

hennishe.DAT

alltafalways

þóttthought

ÓlafurOlaf.NOM

leibinlegur?8

boring.NOM(Icelandic)

‘Has she always considered Olaf boring?’b. Konunginum

the.king.DATvoruwere

gefnargiven.F.PL

ambáttir.9

maidservants.NOM.F.PL‘The king was given female slaves.’

And the approach to agreement also works for cases of remote passive in Ger-man, where the subject is not the first element in an ARG-ST list. One of theexamples in (102) on p. 167 is repreated below:

(25) a. Siethey

erlaubenpermitted

unsus.DAT

nicht,not

denthe.ACC

Erfolgsuccess


‘They did not permit us to enjoy the success.’b. Der

theErfolgsuccess.NOM

wurdewas

unsus.DAT

nichtnot

auszukostento.enjoy

erlaubt.10

permitted‘We were not permitted to enjoy our success.’

See p. 168 for the respective ARG-ST lists.As far as the position in the clause is concerned, the expletive is a subject in

Danish. This is exactlywhatwewant andwhat follows from the generalmappingfrom ARG-ST to SPR and COMPS in SVO languages. The analysis of (26) is shownin Figure 8.9.

8Zaenen, Maling & Thráinsson (1985: 451)9Zaenen, Maling & Thráinsson (1985: 460)10Haider (1986b: 110)



(26) hvemwho

detEXPL

læserreads

bogenbook.DEF


The lexical item for læser ‘to read’ is given in (27):

(27) Lexical item for læser ‘to read’ with expletive subject:SPR

⟨NP[lnom]expl

⟩COMPS ⟨ NP[str], NP[str] ⟩ARG-ST

⟨NP[lnom]expl , NP[str], NP[str]

⟩

The case assignment principles assign nominative to the first NP with structuralcase and accusative to the second. As Figure 8.9 shows, the expletive subject isrealized as specifier in the subject position and the nominative and accusative onthe COMPS list are realized as objects. The “nominative object” is extracted andrealized as the interrogative pronoun.

S

NP[snom]

hvemwho

S/NP[snom]

NP[lnom]

detEXPL

VP/NP[snom]

V′/NP[snom]

V

læserreads

NP[snom]/NP[snom]

_

NP[sacc]

bogenbook.DEF

Figure 8.9: Analysis of interrogative clauses in Danish with subject extraction

Similarly, the analysis of interrogatives in Yiddish may involve an initial ex-pletive in the V2 clause if the speaker finds the subject or any other elementinappropriate for this position for information structural reasons. Figure 8.10shows the analysis of (28):


8.2 The analysis

(28) verwho

esEXPL

leyentreads

dosthe

bukhbook

(Yiddish)


S

NP[snom]

ver𝑖who

S/NP[snom]

NP[lnom]

es𝑗EXPL

S/NP[snom]/NP[lnom]

V ⟨S//V⟩

V

leyent𝑘reads

S//V/NP[snom]/NP[lnom]

NP[lnom]/NP[lnom]

_𝑗

VP//V/NP[snom]

V′//V

V//V

_𝑘

NP[snom]/NP[snom]

_𝑖

NP[sacc]

dos bukhthe book

Figure 8.10: Analysis of a Yiddish interrogative clause involving a fronted exple-tive pronoun

The analysis is more complex than the Danish one, but this is due to the factthat Yiddish has V2 sentences in interrogatives involving verb movement in ad-dition to extraction. Interrogatives have two extracted elements: one for V2, theexpletive in the example, and another one which is the interrogative phrase (ver‘who’ in the example). The figure shows two elements after a /. In the notationusing the SLASH feature, there would be a list with two elements.

This completes the analysis of finite interrogative clauses with and withoutexpletives, but there is more to be said about expletives in general. I stipulated alexical rule adding an expletive element above and this accounts for expletives inYiddish interrogatives. It not just works for interrogatives but for V2 in general.Yiddish declarative V2 clauses can have initial expletives as well, as can German



V2 clauses. (29a) is a German example. As (29b) shows, the expletive is notallowed to appear in the Mittelfeld.

(29) a. EsEXPL

lachenlaugh

dreithree

Kinder.children

‘Three children laugh.’b. * dass

thatesEXPL

dreithree

Kinderchildren

lachenlaugh

In German grammars, this expletive is called “positional es” and it is emphasizedthat it is not the subject and not an argument of the verb (Eisenberg 2004: 129,177, 371; Duden 2005: §1263). The fact that the es cannot appear in the Mittelfeldis seen as support for the non-argumenthood of it. However, we have seen thatthe expletive is realized in the subject position in Danish, so there is some appealto the idea to treat it uniformly as the initial element of the ARG-ST list across theGermanic languages. Nevertheless it is undeniable that the Vorfeld is the onlyplace in which this expletive can appear in German and Yiddish.The problem canbe solved by adding the following constraint to the Expletive Insertion LexicalRule in the grammars of German and Yiddish:

(30) Constraint on the output of the Expletive Insertion Lexical Rule forGerman and Yiddish:[ARG-ST

⟨NP 1 [SLASH ⟨ 1 ⟩]

⟩⊕

]This constraint says that the first element in the ARG-ST list (the expletive) hasto have a SLASH element with the relevant properties of the expletive. Since theexpletive pronoun does not have anything in SLASH, it cannot be combined di-rectly with the respective lexical items. The trace has something in SLASH, this isits very nature. So a trace can combine with the lexical item for lachen ‘to laugh’and then the expletive can function as the filler.

One problem remains: extraction of expletives must be clause-bound:

(31) * Es𝑖EXPL

glaubebelieve

ich,I

dassthat

_𝑖 dreithree

Kinderchildren

lachen.laugh

(German)

Intended: ‘I believe that three children laugh.’

While extraction may cross clause boundaries in principle (see (10) on p. 175), thisis excluded in (31). However, this is not a particular problem of the analysis ofthe positional es at hand but it is a general property of expletive elements. (32)shows an example with the weather es, which clearly is an argument of the verbregnen ‘to rain’:


8.3 Summary

(32) * Es𝑖EXPL

glaubebelieve

ich,I

dassthat

_𝑖 regnet.rains

Intended: ‘I believe that it rains.’

So, whatever rules out examples like (32) also accounts for (31).Finally, there is one problem left: I provided a lexical rule that licenses lexical

item for interrogatives with an expletive in subject position, but what is stillmissing is a constraint that rules out sentences without the expletive. There isnothing in the grammar so far that does this. It is possible to formulate somethinglike this but the formal tools have not been introduced in this book. The readeris refered to Müller & Ørsnes (2011: 185) for details.

8.3 Summary

This chapter provided an analysis of dependent clauses introduced by a comple-mentizer and of interrogative clauses. Together with the V1 and V2 clauses dealtwith in Chapter 6 this covers the main clause types in the Germanic languages.The variation in these subordinated clauses is connected to what we saw be-fore: the SOV languages have SOV order in embedded languages and some SVOlanguages have SVO order, some allow for both SVO and V2 and some allowfor V2 only. Interrogative clauses involve a clause with a gap and the filler isthe interrogative phrase containing a wh-word in English and a correspondingword in the other Germanic languages. The wh phrase may consist of a singleinterrogative pronoun or may be internally complex. The information about theinterrogative pronoun has to be present at the top-most node of the interrogativephrase for semantic and syntactic reasons. The syntactic reason is of course thatone has to make sure that the fronted phrase contains an interrogative pronounat all. The information is passed up from the interrogative pronoun by the samemechanism that is also used for extraction: like SLASH, qUE is used to pass theinformation up. Danish and Yiddish use expletive pronouns in interrogatives. Toaccount for this, Müller & Ørsnes (2011) suggested a lexical rule that introducedthe expletive into the ARG-ST list. This expletive can function as subject in Danishand as positional expletive in German and Yiddish.


9 Outlook

This book sketched fragments of grammars of several Germanic languages. Thetheory lurking in the background is Head-Driven Phrase Structure Grammar(HPSG) (Pollard & Sag 1987; 1994; Müller 2013b; Müller et al. 2021). We had alook at valence and how it is represented in valence lists like SPR and COMPS.We also looked at adjuncts, which are not represented in lists: adjuncts selectthe heads they modify. HPSG assumes that there are schemata for the com-binaiton of linguistic material. We dealt with the Specifier-Head Schema, theHead-Complements Schema, the Head-Adjunct Schema and also the PredicateComplex Schema. Verbal complexes in the Germanic OV languages have beenanalyzed as predicate complex formation.

The Germanic languages vary as far as their basic order is concerned (VO orOV). Apart from English all Germanic languages are V2 languages. V2 sentencesare analyzed via head movement: there is an empty verb in final position whichis related to the fronted verb.

All analyses are implemented in computer-processable grammar fragments.They are fully formalized – otherwise they would not be processable – but theyhave been given here in simplified and sketchy form. I briefly talked about theconnection between syntax and semantics in Section 4.8, but of course all imple-mentations come with semantic representations.

Due to space limitations, it is not possible to carefully introduce all concepts ofHPSG, but the interested reader is invited to have a look at theHPSGmonographs(Pollard & Sag 1987; 1994; Ginzburg & Sag 2000; Müller 2013b), overview articles(Levine & Meurers 2006; Przepiórkowski & Kupść 2006; Bildhauer 2014; Müller2015a; Müller & Machicao y Priemer 2019), the respective chapter in the Gram-matical Theory textbook (Müller 2020: Chapter 9) or the handbook on HPSG(Müller et al. 2021). Especially the latter volume is an uptodate book with morethan 1600 pages dealing in 32 chapters with almost every aspect one could beinterested in.

Appendix A: Solutions

A.1 Phrase structure grammars and X theory

1. Draw trees for the following phrases. You may use the symbol NP forproper names andN for nouns not requiring complements (as in Figure 3.11).

(1) a. eineone

Stundehour

vorbefore

derthe

Ankunftarrival

desof.the

Zugestrain


shortlynachafter

derthe

Ankunftarrival

inin

ParisParis


thiseina

Liedsong

singendesinging

Kindchild

ausfrom

demthe

AllgäuAllgäu


A Solutions

PP

NP

Det

eineone

N

Stundehour

P

P

vorbefore

NP

Det

derthe

N

N

Ankunftarrival

NP

Det

desthe

N

Zugestrain

Figure A.1: Analysis of eine Stunde vor der Ankunft des Zuges ‘one hour beforethe arrival of the train’


A.1 Phrase structure grammars and X theory

PP

AP

kurzshortly

P

P

nachafter

NP

Det

derthe

N

N

Ankunftarrival

PP

P

P

inin

NP

ParisParis

Figure A.2: Analysis of kurz nach der Ankunft in Paris ‘shortly after the arrival inParis’


A Solutions

NP

Det

dasthe

N

N

AP

A

NP

Det

eina

N

Liedsong

A

singendesinging

N

Kindchild

PP

P

P

ausfrom

NP

Det

demthe

N

AllgäuAllgäu

Figure A.3: Analysis of das ein Lied singende Kind aus dem Allgäu ‘the child fromthe Allgäu singing a song’


A.2 Valency, argument order and adjunct placement


1. Provide the valency lists for the following words:

(2) a. laugh SPR ⟨ NP[nom] ⟩b. eat SPR ⟨ NP[nom] ⟩, COMPS ⟨ NP[acc] ⟩c. to douse SPR ⟨ NP[nom] ⟩, COMPS ⟨ NP[acc] ⟩d. bezichtigen SPR ⟨⟩, COMPS

⟨NP[nom], NP[gen]

⟩(German)

accusee. he SPR ⟨⟩, COMPS ⟨⟩f. the SPR ⟨⟩, COMPS ⟨⟩g. Ankunft SPR ⟨ DET ⟩, COMPS ⟨ NP ⟩ (German)

arrival SPR ⟨ DET ⟩, COMPS ⟨⟩

If you are uncertain as far as case assignment is concerned, you may usethe Wiktionary1.

2. Draw trees for the NPs that were also used in exercise 1 on page 60 inChapter 3.

(3) a. eineone

Stundehour

vorbefore

derthe

Ankunftarrival

desof.the

Zugestrain


shortlynachafter

derthe

Ankunftarrival

inin

ParisParis


thiseina

Liedsong

singendesinging

Kindchild

ausfrom

demthe

AllgäuAllgäu


3. Draw trees for the following examples. NPs can be abbreviated.

(4) a. weilbecause

AickeAicke

demthe.DAT

Kindchild

eina.ACC

Buchbook

schenktgives.as.a.present

(German)

‘because Aicke gives the child a book as a present’

1https://de.wiktionary.org/, 2018-07-02.



A Solutions

P[SPR ⟨⟩,COMPS ⟨⟩]

1 N[SPR ⟨⟩,COMPS ⟨⟩]

2 Det

eineone

N[SPR ⟨ 2 ⟩,COMPS ⟨⟩]

Stundehour

P[SPR ⟨ 1 ⟩,COMPS ⟨⟩]

P[SPR ⟨ 1 ⟩,COMPS ⟨ 3 ⟩]

vorbefore


4 Det

derthe


N[SPR ⟨ 4 ⟩,COMPS ⟨ 5 ⟩]

Ankunftarrival


6 Det

desthe


Zugestrain

Figure A.4: Analysis of eine Stunde vor der Ankunft des Zuges ‘one hour beforethe arrival of the train’

b. weilbecause

demthe.DAT

Kindchild

solchsuch

eina.ACC

Buchbook

niemandnobody.NOM

schenktgives.as.a.present‘because nobody gives the child such a book as a present’

c. because Kim gave a book to himd. Sandy saw this yesterday.e. at

thatBjarneBjarne

læsteread

bogenbook.DEF

(Danish)

‘that Bjarne read the book’

The trees with the solutions are given in the following. Figure A.8 differs



P[SPR ⟨⟩,COMPS ⟨⟩]

1 A[SPR ⟨⟩,COMPS ⟨⟩]

kurzshortly

P[SPR ⟨ 1 ⟩,COMPS ⟨⟩]

P[SPR ⟨ 1 ⟩,COMPS ⟨ 2 ⟩]

nachafter


3 Det

derthe


4 N[SPR ⟨ 3 ⟩,COMPS ⟨⟩]

Ankunftarrival

P[MOD 4 ,SPR ⟨⟩,COMPS ⟨⟩]

P[MOD 4 ,SPR ⟨⟩,COMPS ⟨ 5 ⟩]

inin


ParisParis

Figure A.5: Analysis of kurz nach der Ankunft in Paris ‘shortly after the arrival inParis’

from Figure A.7 in the way the elements in the COMPS list are numbered,but in each case the order of the elements in the COMPS list of schenkt ‘givesas a present’ is ⟨ NP[nom], NP[dat], NP[acc]⟩. The different numbering isdue to the order in which the elements are combined. If the numberingis done consistently from top to bottom, Figure A.8 is the result. If one ismore liberal in the way the numbers are assigned, the same situation canbe depicted as in Figure A.9. Figure A.9 has the same numbering in thevalence list a Figure A.7 and maybe easier to grasp because of this.


A Solutions

N[SPR ⟨⟩,COMPS ⟨⟩]

1 Det

dasthe


2 N[SPR ⟨ 1 ⟩,COMPS ⟨⟩]

A[SPR ⟨⟩,COMPS ⟨⟩]


6 Det

eina


Liedsong

A[SPR ⟨⟩ ,COMPS ⟨ 4 ⟩]

singendesinging


Kindchild

P[MOD 2 ,SPR ⟨⟩,COMPS ⟨⟩]

P[MOD 2 ,SPR ⟨⟩,COMPS ⟨ 3 ⟩]

ausfrom


5 Det

demthe


AllgäuAllgäu

Figure A.6: Analysis of das ein Lied singende Kind aus dem Allgäu ‘the child fromthe Allgäu singing a song’



C[SPR ⟨⟩, COMPS ⟨ ⟩]

C[SPR ⟨⟩, COMPS ⟨ 1 ⟩]

weilbecause


2 NP[nom]

AickeAicke

V[SPR ⟨ ⟩, COMPS ⟨ 2 ⟩]

3 NP[dat]

dem Kindthe child

V[SPR ⟨ ⟩, COMPS ⟨ 2 , 3 ⟩]

4 NP[acc]

ein Bucha book

V[SPR ⟨ ⟩,COMPS ⟨ 2 , 3 , 4 ⟩]

schenktgives

Figure A.7: The analysis of weil Aicke dem Kind ein Buch schenkt ‘because Aickegives the child a book as a present’



weilbecause


2 NP[acc]

solch ein Buchsuch a book

V[SPR ⟨ ⟩, COMPS ⟨ 2 ⟩]

3 NP[dat]

dem Kindthe child

V[SPR ⟨ ⟩, COMPS ⟨ 3 , 2 ⟩]

4 NP[nom]

niemandnobody

V[SPR ⟨ ⟩,COMPS ⟨ 4 , 3 , 2 ⟩]

schenktgives

Figure A.8: The analysis of weil dem Kind solch ein Buch niemand schenkt ‘be-cause nobody gives the child such a book as a present’


A Solutions



weilbecause


4 NP[acc]

solch ein Buchsuch a book

V[SPR ⟨ ⟩, COMPS ⟨ 4 ⟩]

3 NP[dat]

dem Kindthe child

V[SPR ⟨ ⟩, COMPS ⟨ 3 , 4 ⟩]

2 NP[nom]

niemandnobody

V[SPR ⟨ ⟩,COMPS ⟨ 2 , 3 , 4 ⟩]

schenktgives

Figure A.9: The analysis of weil dem Kind solch ein Buch niemand schenkt ‘be-cause nobody gives the child such a book as a present’



because


2 NP[nom]

Kim

V[SPR ⟨ 2 ⟩, COMPS ⟨ ⟩]

V[SPR ⟨ 2 ⟩, COMPS ⟨ 3 ⟩]

V[SPR ⟨ 2 ⟩,COMPS ⟨ 4 , 3 ⟩]

gave

4 NP[acc]

a book

3 PP[to]

to him

Figure A.10: The analysis of because Kim gave a book to him




1 NP[nom]

Sandy

V[SPR ⟨ 1 ⟩, COMPS ⟨⟩]

2 V[SPR ⟨ 1 ⟩, COMPS ⟨⟩]

V[SPR ⟨ 1 ⟩,COMPS ⟨ 3 ⟩]

saw

3 NP[acc]

this

Adv[MOD 2 VP]

yesterday

Figure A.11: Analysis of Sandy saw this yesterday.



atthat


2 NP[nom]

BjarneBjarne

V[SPR ⟨ 2 ⟩, COMPS ⟨⟩]

V[SPR ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

læsteread

3 NP[acc]

bogenbook.DEF

Figure A.12: Analysis of at Bjarne læste bogen ‘that Bjarne read the book’


A Solutions

A.3 The verbal complex

1. Sketch the analysis of the verbal complexes in the following examples:

(5) a. dassthat

sieshe

darüberthere.about

lachenlaugh

mussmust

(German)

‘that she has to laugh about it’b. dass

thatsieshe

darüberthere.about

hathas

lachenlaugh

müssenmust

‘that she had to laugh about it’c. dass

thatsieshe

darüberthere.about

wirdwill

habenhave

lachenlaugh

müssenmust

‘that it will be the case that she had to laugh about it’

You may omit the SPR values, since they are the empty list for all Germanverbs anyway.

C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

dassthat

1 V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V[COMPS ⟨ 2 , 3 ⟩]

4 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

lachenlaugh

V[COMPS ⟨ 2 , 3 , 4 ⟩]

mussmust

Figure A.13: Analysis of dass sie darüber lachen muss ‘that she has to laugh aboutthis’


A.3 The verbal complex

C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

dassthat

1 V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V[COMPS ⟨ 2 , 3 ⟩]

V[COMPS ⟨ 2 , 3 , 4 ⟩]

hathas

4 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

5 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

lachenlaugh

V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 , 5 ⟩]

müssenmust

Figure A.14: Analysis of dass sie darüber hat lachen müssen ‘that she had to laughabout this’


A Solutions

C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

dassthat

1 V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V[COMPS ⟨ 2 , 3 ⟩]

V[COMPS ⟨ 2 , 3 , 4 ⟩]

wirdwill

4 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 , 5 ⟩]

habenhave

5 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

6 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

lachenlaugh

V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 , 6 ⟩]

müssenmust

Figure A.15: Analysis of dass sie darüber wird haben lachen müssen ‘that it will bethe case that she had to laugh about this’


A.4 Verb position: Verb first and verb second


1. Classify the Germanic languages according to their basic constituent or-der (SVO, SOV, VSO, …) and V2 assuming that you know that one of thefollowing patterns exist in the language:

(6) a. NP[acc] V-Aux NP[nom] V NP[dat] V2 SVOb. NP[acc] V-Aux NP[nom] NP[dat] V V2 SOVc. NP[acc] NP[nom] V NP[acc] −V2 SVOd. NP[acc] NP[nom] V-Aux V NP[acc] −V2 SVOe. NP[acc] V-Aux NP[nom] V PP not classifiable

The pattern in (6a) cannot be Englih, since English does not have a dative.Hence it is a V2 language. The dative object follows the verb, so it must bean SVO language. An example would be Icelandic.

(7)

(6b) has an auxiliary and two NPs followed by a verb. Since the dative ob-ject would follow the verb in an SVO language, it must be a SOV language.Since all Germanic SOV languages are also V2 languages, (6b) must be aV2 language. German and Dutch would be examples.

(8) Denthe.ACC

Romannovel

hathas

jemandsomebody.NOM

demthe.DAT

Kindchild

gegeben.given

‘Somebody has given the child the novel.’

Ignoring multiple frontings in German (Müller 2003a), (6c) must be a non-V2 pattern. The language can only be English:

(9) This book, Kim gave Sandy.

For the same reason, (6d) is non-V2 and SVO. The language must be En-glish:

(10) This book, Kim had given Sandy.

The pattern in (6e) cannot be unambiguously classified with respect to V2and SOV/SVO. Since PPs can be extraposed easily, it could be an SOV lan-gauge with extraposition (e.g., German) or it could be English with ques-tion formation (residual V2):


A Solutions

(11) a. Wenwho

hathas

AickeAicke

gesehenseen

beiduring

derthe

Demonstration.rally

‘Who has Aicke seen during the rally.’b. Who did Kim see during the rally?

2. Sketch the analysis for the following examples. Use the abbreviations usedin this chapter, that is, do not go into the details regarding SPR and COMPSvalues but use S, VP, and V′.


BjarneBjarne

ihærdigtseriously

påat

bogen?book.DEF

(Danish)


worksBjarneBjarne

ernsthaftseriously

anat

demthe

Buch?book

(German)

‘Does Bjarne work seriously on the book?’c. Wird

willsieshe

darüberthere.upon

nachdenken?think

(German)


S

V⟨ S//V ⟩

V

arbejderworks

S//V

NP

BjarneBjarne

VP//V

Adv

ihærdigtseriously

VP//V

V//V

_

PP

på bogenat book.DEF

Figure A.16: Analysis of Arbejder Bjarne ihærdigt på bogen? ‘Does Bjarne workseriously on the book?’



S

V⟨ S//V ⟩

V

arbeitetworks

S//V

NP

BjarneBjarne

V′//V

Adv

ernsthaftseriously

V′//V

PP

an dem Buchat the book

V//V

_


S

V[⟨ S//V ⟩]

V

wirdwill

S//V

NP

sieshe

V′//V

PP

darüberthere.about

V//V

V


V//V

_

Figure A.18: Analysis of Wird sie darüber nachdenken? ‘Will she think aboutthis?’


A Solutions

3. Sketch the analysis for the following examples. Use the valence featuresSPR and COMPS rather than the abbreviations S, VP, and V′. Since the valueof SPR in German is always the empty list, you may omit it in the Germanexamples. NPs and PPs can be abbreviated as NP and PP, respectively.

(13) a. dassthat

sieshe

darüberthere.upon


(German)

‘that she laughs about this’b. dass

thatsieshe

darüberthere.upon


wirdwill

‘that she will think about this’c. Wird

willsieshe

darüberthere.upon

nachdenken?think



BjarneBjarne

ihærdigtseriously

påat

bogen?book.DEF

(Danish)


worksBjarneBjarne

ernsthaftseriously

anat

demthe

Buch?book

(German)

‘Does Bjarne work seriously on the book?’

C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

dassthat

1 V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V[COMPS ⟨ 2 , 3 ⟩]


Figure A.19: Analysis of dass sie darüber nachdenkt ‘that she thinks about this’



C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

dassthat

1 V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V[COMPS ⟨ 2 , 3 ⟩]

4 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]


V[COMPS ⟨ 2 , 3 , 4 ⟩]

wirdwill

Figure A.20: Analysis of dass sie darüber nachdenken wird ‘that she will thinkabout this’

V[COMPS ⟨ ⟩]

V[COMPS ⟨ 1 ⟩]

V

wirdwill

1 V//V[COMPS ⟨ ⟩]

2 NP[nom]

sieshe

V//V[COMPS ⟨ 2 ⟩]

3 PP

darüberthere.about

V//V[COMPS ⟨ 2 , 3 ⟩]

4 V[SUBJ ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]


V//V[COMPS ⟨ 2 , 3 , 4 ⟩]

_

Figure A.21: Analysis of Wird sie darüber nachdenken? ‘Will she think aboutthis?’


A Solutions

V[COMPS ⟨ ⟩]

V[COMPS ⟨ 1 ⟩]

V

arbejderworks

1 V//V[SPR ⟨⟩,COMPS ⟨⟩]

2 NP

BjarneBjarne

V//V[SPR ⟨ 2 ⟩,COMPS ⟨⟩]

Adv[MOD 3 VP]

ihærdigtseriously

3 V//V[SPR ⟨ 2 ⟩,COMPS ⟨⟩]

V//V[SPR ⟨ 2 ⟩,COMPS ⟨ 4 ⟩]

_

4 PP

på bogenat book.DEF


V[COMPS ⟨ ⟩]

V[COMPS ⟨ 1 ⟩]

V

arbeitetworks

1 V//V[COMPS ⟨⟩]

2 NP

BjarneBjarne

V//V[COMPS ⟨ 2 ⟩]

Adv[MOD 3 V[INI−]]

ernsthaftseriously

3 V//V[COMPS ⟨ 2 ⟩]

4 PP

an dem Buchat the book

V//V[COMPS ⟨ 2 , 4 ⟩]

_

Figure A.23: Analysis of Arbeitet Bjarne ernsthaft an dem Buch? ‘Does Bjarnework seriously on the book?’



4. Sketch the analysis of the following examples. NPs may be abbreviated.Valence features should not be given, but node labels like V, V′, VP andS should be used instead. If non-local dependencies are involved indicatethem using the ‘/’ symbol.

(15) a. Such books, I like.b. Solche

suchBücherbooks

maglike

ich.I

(German)

‘I like such books.’c. Boger

bookssomlike

detthis

elskerlike

jeg.I

(Danish)

‘I like such books.’

S

NP𝑖

such books

S/NP

NP

I

VP/NP

V

like

NP/NP

_𝑖

Figure A.24: Analysis of Such books, I like.


A Solutions

S

NP𝑖

solche Büchersuch books

S/NP

V⟨S//V

⟩V

mag𝑗like

S//V/NP

NP/NP

_𝑖

V′//V

NP

ichI

V//V

_𝑗

Figure A.25: Analysis of the German example Solche Bücher mag ich. ‘I like suchbooks.’

S

NP𝑖

boger som detbooks like this

S/NP

V⟨S//V

⟩V

elsker𝑗like

S//V/NP

NP

jegI

VP//V/NP

V//V

_𝑗

NP/NP

_𝑖

Figure A.26: Analysis of the Danish example Boger som det elsker jeg. ‘I like suchbooks.’


A.5 Passive

A.5 Passive

C[COMPS ⟨ ⟩]

C[COMPS ⟨ 1 ⟩]

that

1 V[SPR ⟨ ⟩,COMPS ⟨ ⟩]

2 NP[nom]

the box

V[SPR ⟨ 2 ⟩,COMPS ⟨ ⟩]

V[SPR ⟨ 2 ⟩,COMPS ⟨ 3 ⟩]

was

3 V[SPR ⟨ 2 ⟩,COMPS ⟨⟩]

opened

Figure A.27: Analysis of the passive sentences that the box was opened

The transitive verb open takes a subject and an object. The ARG-ST list containstwo NPs with structural case. The passive lexical rule removes one argument.For the passive participle this leaves us with one element on the ARG-ST list. Thiselement gets mapped to the SPR list of opened. The passive auxiliary takes a VP inpassive form and takes over its element from SPR. After combination of auxiliaryand passive VP, we have the VP was opened still selecting for a specifier. The NPthe box functions as the specifier and the combination of the box and was openedis a complete sentence.

A.6 Clause types and expletives


References

Ackerman, Farrell, Robert Malouf & John Moore. 2013. Symmetric objects inMoro. Paper presentend at the 20th International Conference on Head-DrivenPhrase Structure Grammar, Freie Universität Berlin. http://csli-publications.stanford.edu/HPSG/2013/abstr-amm.pdf (22 September, 2020).

Adams, Marianne. 1987. From Old French to the theory of pro-drop. Natural Lan-guage & Linguistic Theory 5(1). 1–32. DOI: 10.1007/BF00161866.

Allwood, Jens, Lars-Gunnar Anderson & Östen Dahl. 1973. Logik für Linguisten(Romanistische Arbeitshefte 8). Tübingen: Max Niemeyer Verlag.

Alsina, Alex. 1996. Resultatives: A joint operation of semantic and syntactic struc-tures. In Miriam Butt & Tracy Holloway King (eds.), Proceedings of the LFG ’96conference, Rank Xerox, Grenoble, 1–15. Stanford, CA: CSLI Publications. http://csli-publications.stanford.edu/LFG/1/ (10 February, 2021).

Anderson, Stephen R. 2006. Verb second, subject clitics, and impersonals in Sur-miran (Rumantsch). In Zhenya Antić, Charles B. Chang, Emily Cibelli, JisupHong, Michael J. Houser, Clare S. Sandy, Maziar Toosarvandani & Yao Yao(eds.), Proceedings of the 32nd Annual Meeting of the Berkeley Linguistics Soci-ety: Theoretical approaches to argument structure, vol. 32, 3–21. Berkeley, CA:Berkeley Linguistics Society. DOI: 10.3765/bls.v32i1.3435.

Askedal, John Ole. 1988. Über den Infinitiv als Subjekt im Deutschen. Zeitschriftfür Germanistische Linguistik 16(1). 1–25. DOI: 10.1515/zfgl.1988.16.1.1.

Asudeh, Ash, Mary Dalrymple & Ida Toivonen. 2008. Constructions with lexicalintegrity: Templates as the lexicon-syntax interface. In Miriam Butt & TracyHolloway King (eds.), Proceedings of the LFG ’08 conference, University of Syd-ney, 68–88. Stanford, CA: CSLI Publications. http://csli-publications.stanford.edu/LFG/13/papers/lfg08asudehetal.pdf (10 February, 2021).

Asudeh, Ash, Mary Dalrymple & Ida Toivonen. 2013. Constructions with lexicalintegrity. Journal of Language Modelling 1(1). 1–54. DOI: 10.15398/jlm.v1i1.56.

Bartsch, Renate & Theo Vennemann. 1972. Semantic structures: A study in the re-lation between semantics and syntax (Athenäum-Skripten Linguistik 9). Frank-furt/Main: Athenäum.

http://csli-publications.stanford.edu/HPSG/2013/abstr-amm.pdf

http://csli-publications.stanford.edu/HPSG/2013/abstr-amm.pdf

https://doi.org/10.1007/BF00161866

http://csli-publications.stanford.edu/LFG/1/


https://doi.org/10.3765/bls.v32i1.3435

https://doi.org/10.1515/zfgl.1988.16.1.1

http://csli-publications.stanford.edu/LFG/13/papers/lfg08asudehetal.pdf

http://csli-publications.stanford.edu/LFG/13/papers/lfg08asudehetal.pdf

https://doi.org/10.15398/jlm.v1i1.56

References

Bech, Gunnar. 1955. Studien über das deutsche Verbum infinitum (Historisk-filologiske Meddekekser udgivet af Det Kongelige Danske Videnskabernes Sel-skab. Bind 35, no. 2, 1955; Bind 36, no. 6, 1957). København: Det KongeligeDanske Videnskabernes Selskab. Reprint as: Studien über das deutsche Ver-bum infinitum, 2nd edn. 1983 (Linguistische Arbeiten 139). Tübingen: Max Nie-meyer Verlag, 1983.

Behaghel, Otto. 1932. Die deutsche Syntax: Eine geschichtliche Darstellung: Wort-stellung. Periodenbau. Vol. 4 (Germanische Bibliothek 1: Sammlung german-ischer Elementar- und Handbücher. R. 1: Grammatiken 10). Heidelberg: CarlWinters Universitätsbuchhandlung.

Berman, Judith. 2003.Clausal syntax of German (Studies in Constraint-Based Lex-icalism). Stanford, CA: CSLI Publications.

Bhatt, Rakesh Mohan. 1999. Verb movement and the syntax of Kashmiri (Studiesin Natural Language and Linguistic Theory 46). Dordrecht: Kluwer AcademicPublishers. DOI: 10.1007/978-94-015-9279-6.

Bierwisch, Manfred. 1963. Grammatik des deutschen Verbs (studia grammatica 2).Berlin: Akademie Verlag.

Bildhauer, Felix. 2014. Head-Driven Phrase Structure Grammar. In AndrewCarnie, Yosuke Sato & Dan Siddiqi (eds.), The Routledge handbook of syntax,526–555. Oxford: Routledge. DOI: 10.4324/9781315796604.

Bjerre, Anne & Tavs Bjerre. 2007. Perfect and periphrastic passive construc-tions in Danish. Nordic Journal of Linguistics 30(1). 5–53. DOI: 10 . 1017 /S0332586507001643.

Blackburn, Patrick & Johan Bos. 2005. Representation and inference for naturallanguage: A first course in computational semantics (Studies in ComputationalLinguistics). Stanford, CA: CSLI Publications.

Bloomfield, Leonard. 1933. Language. New York, NY: Holt, Rinehart, & Winston.Bonami, Olivier, Danièle Godard & B. Kampers-Manhe. 2004. Adverb classifica-

tion. In Francis Corblin &Henriëtte de Swart (eds.),Handbook of French seman-tics (CSLI Lecture Notes 170), 143–184. Stanford, CA: CSLI Publications.

Borsley, Robert D. 1987. Subjects and complements in HPSG. Report No. CSLI-87-107. Stanford, CA: Center for the Study of Language & Information.

Borsley, Robert D. & Berthold Crysmann. 2021. Unbounded dependencies. In Ste-fan Müller, Anne Abeillé, Robert D. Borsley & Jean-Pierre Koenig (eds.), Head-Driven Phrase Structure Grammar: The handbook (Empirically Oriented Theo-retical Morphology and Syntax). To appear. Berlin: Language Science Press.

Borsley, Robert D. & Andreas Kathol. 2000. Breton as a V2 language. Linguistics38(4). 665–710. DOI: 10.1515/ling.2000.002.


https://doi.org/10.1007/978-94-015-9279-6

https://doi.org/10.4324/9781315796604

https://doi.org/10.1017/S0332586507001643

https://doi.org/10.1017/S0332586507001643

https://doi.org/10.1515/ling.2000.002

Borsley, Robert D., Maggie Tallerman & David Willis. 2007. The syntax of Welsh(Cambridge Syntax Guides 7). Cambridge: Cambridge University Press. DOI:10.1017/CBO9780511486227.

Bouma, Gosse. 2003. Verb clusters and the scope of adjuncts in Dutch. In PieterA. M. Seuren & Gerard Kempen (eds.), Verb constructions in German and Dutch(Current Issues in Linguistic Theory 242), 5–42. Amsterdam: John BenjaminsPublishing Co. DOI: 10.1075/cilt.242.02bou.

Bresnan, Joan (ed.). 1982. The mental representation of grammatical relations (MITPress Series onCognitive Theory andMental Representation). Cambridge,MA:MIT Press.

Bresnan, Joan, Ash Asudeh, Ida Toivonen & StephenWechsler. 2016. Lexical-func-tional syntax. 2nd edn. (Blackwell Textbooks in Linguistics 16). Oxford: Wiley-Blackwell. DOI: 10.1002/9781119105664.

Chomsky, Noam. 1957. Syntactic structures (Janua Linguarum / Series Minor 4).Berlin: Mouton de Gruyter. DOI: 10.1515/9783112316009.

Chomsky, Noam. 1970. Remarks on nominalization. In Roderick A. Jacobs & PeterS. Rosenbaum (eds.), Readings in English Transformational Grammar, 184–221.Waltham, MA: Ginn & Company.

Chomsky, Noam. 1981. Lectures on government and binding (Studies in GenerativeGrammar 9). Dordrecht: Foris Publications. DOI: 10.1515/9783110884166.

Christie, Elizabeth. 2010. Using templates to account for English resultatives. InMiriam Butt & Tracy Holloway King (eds.), Proceedings of the LFG ’10 con-ference, Carleton University, Ottawa, 155–164. Stanford, CA: CSLI Publications.http://csli-publications.stanford.edu/LFG/15/ (10 February, 2021).

Copestake, Ann, Dan Flickinger, Carl Pollard & Ivan A. Sag. 2005. Minimal Re-cursion Semantics: An introduction. Research on Language and Computation3(2–3). 281–332. DOI: 10.1007/s11168-006-6327-9.

Crysmann, Berthold. 2004. Underspecification of intersective modifier attach-ment: Some arguments from German. In Stefan Müller (ed.), Proceedings ofthe 11th International Conference on Head-Driven Phrase Structure Grammar,Center for Computational Linguistics, Katholieke Universiteit Leuven, 378–392.Stanford, CA: CSLI Publications. http://csli-publications.stanford.edu/HPSG/2004/crysmann.pdf (10 February, 2021).

Culicover, Peter W. & Ray S. Jackendoff. 2005. Simpler Syntax (Oxford Linguis-tics). Oxford: Oxford University Press. DOI: 10.1093/acprof:oso/9780199271092.001.0001.

Davis, Anthony R. 2001. Linking by types in the hierarchical lexicon (Studies inConstraint-Based Lexicalism 10). Stanford, CA: CSLI Publications.


https://doi.org/10.1017/CBO9780511486227

https://doi.org/10.1075/cilt.242.02bou

https://doi.org/10.1002/9781119105664

https://doi.org/10.1515/9783112316009

https://doi.org/10.1515/9783110884166


https://doi.org/10.1007/s11168-006-6327-9

http://csli-publications.stanford.edu/HPSG/2004/crysmann.pdf

http://csli-publications.stanford.edu/HPSG/2004/crysmann.pdf

https://doi.org/10.1093/acprof:oso/9780199271092.001.0001

https://doi.org/10.1093/acprof:oso/9780199271092.001.0001

References

Davis, Anthony R., Jean-Pierre Koenig & Stephen Wechsler. 2021. Argumentstructure and linking. In Stefan Müller, Anne Abeillé, Robert D. Borsley &Jean-Pierre Koenig (eds.), Head-Driven Phrase Structure Grammar: The hand-book (Empirically Oriented Theoretical Morphology and Syntax). To appear.Berlin: Language Science Press.

De Kuthy, Kordula, Vanessa Metcalf & Walt Detmar Meurers. 2004. Documenta-tion of the implementation of the Milca English Resource Grammar in the Tralesystem. Ohio State University, ms.

den Besten, Hans. 2012. From Khoekhoe foreignertalk via Hottentot Dutch toAfrikaans: The creation of a novel grammar. In Ton van der Wouden (ed.),Roots of Afrikaans: Selected writings of Hans Den Besten (Creole Language Li-brary 44), 257–287. Amsterdam: John Benjamins Publishing Co. DOI: 10.1075/cll.44. Reprint from: Hans den Besten. 1989. From Khoekhoe foreignertalk viaHottentot Dutch to Afrikaans: The creation of a novel grammar. In Wheelswithin wheels. Papers of the Duisburg Symposium on Pidgin and Creole Lan-guages, 207–209. Frankfurt a.M.: Peter Lang, 1989.

Diesing, Molly. 1990. Verb movement and the subject position in Yiddish.NaturalLanguage & Linguistic Theory 8(1). 41–79. DOI: 10.1007/BF00205531.

Dowty, David. 1991. Thematic proto-roles and argument selection. Language67(3). 547–619. DOI: 10.2307/415037.

Dryer, Matthew S. 2013a. Determining dominant word order. In Matthew S.Dryer & Martin Haspelmath (eds.), The world atlas of language structures on-line. Leipzig: Max Planck Institute for Evolutionary Anthropology. https : / /wals.info/chapter/s6 (17 February, 2021).

Dryer, Matthew S. 2013b. Order of subject, object and verb. InMatthew S. Dryer &Martin Haspelmath (eds.), The world atlas of language structures online. Leipzig:Max Planck Institute for Evolutionary Anthropology. http://wals.info/chapter/81 (10 February, 2021).

Duden. 2005. Duden: Die Grammatik. 7th edn. Vol. 4. Mannheim, Leipzig, Wien,Zürich: Dudenverlag.

Eisenberg, Peter. 1994. German. In Ekkehard König & Johan van der Auwera(eds.), The Germanic languages (Routledge Language Family Descriptions 3),349–387. London: Routledge. DOI: 10.4324/9781315812786.

Eisenberg, Peter. 2004. Grundriß der deutschen Grammatik: Der Satz. 2nd edn.Vol. 2. Stuttgart, Weimar: Verlag J. B. Metzler.

Eisenberg, Peter, Jörg Peters, Peter Gallmann, Cathrine Fabricius-Hansen,Damaris Nübling, Irmhild Barz, Thomas A. Fritz & Reinhard Fiehler. 2005. Du-


https://doi.org/10.1075/cll.44

https://doi.org/10.1075/cll.44

https://doi.org/10.1007/BF00205531

https://doi.org/10.2307/415037

https://wals.info/chapter/s6

https://wals.info/chapter/s6

http://wals.info/chapter/81

http://wals.info/chapter/81

https://doi.org/10.4324/9781315812786

den: Die Grammatik. 7th edn. Vol. 4. Mannheim, Leipzig, Wien, Zürich: Duden-verlag.

Fanselow, Gisbert. 2001. Features, θ-roles, and free constituent order. LinguisticInquiry 32(3). 405–437. DOI: 10.1162/002438901750372513.

Fanselow, Gisbert. 2003. Münchhausen-style head movement and the analysisof verb second. In Anoop Mahajan (ed.), Proceedings of the workshop on headmovement (UCLA Working Papers in Linguistics 10). Los Angeles: UCLA, Lin-guistics Department.

Fillmore, Charles J., Russell R. Lee-Goldmann & Russell Rhomieux. 2012. TheFrameNet constructicon. In Hans C. Boas & Ivan A. Sag (eds.), Sign-Based Con-struction Grammar (CSLI Lecture Notes 193), 309–372. Stanford, CA: CSLI Pub-lications.

Fitch, W. Tecumseh. 2007. Linguistics: An invisible hand. Nature 449(11). 665–667.Fontana, Josep M. 1997. On the integration of second position phenomena. In

Parameters of morphosyntactic change, 207–249. Cambridge, UK: CambridgeUniversity Press.

Frey, Werner. 1993. Syntaktische Bedingungen für die semantische Interpretation:Über Bindung, implizite Argumente und Skopus (studia grammatica 35). Berlin:Akademie Verlag.

Frey, Werner. 2004. The grammar-pragmatics interface and the German prefield.Forschungsprogramm Sprache und Pragmatik 52. Germanistisches Institut derUniversität Lund. 1–39.

Fries, Norbert. 1988. Über das Null-Topik im Deutschen. ForschungsprogrammSprache und Pragmatik 3. Germanistisches Institut der Universität Lund. 19–49.

Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum & Ivan A. Sag. 1985. GeneralizedPhrase Structure Grammar. Cambridge, MA: Harvard University Press.

Gazdar, Gerald, Geoffrey K. Pullum & Ivan A. Sag. 1982. Auxiliaries and relatedphenomena in a restrictive theory of grammar. Language 58(3). 591–638. DOI:10.2307/413850.

Geach, Peter Thomas. 1970. A program for syntax. Synthese 22(1–2). 3–17. DOI:10.1007/BF00413597.

Ginzburg, Jonathan & Ivan A. Sag. 2000. Interrogative investigations: The form,meaning, and use of English interrogatives (CSLI Lecture Notes 123). Stanford,CA: CSLI Publications.

Godard, Danièle & Pollet Samvelian. 2021. Complex predicates. In Stefan Müller,Anne Abeillé, Robert D. Borsley & Jean-Pierre Koenig (eds.), Head-Driven


https://doi.org/10.1162/002438901750372513

https://doi.org/10.2307/413850

https://doi.org/10.1007/BF00413597

References

Phrase Structure Grammar: The handbook (Empirically Oriented TheoreticalMorphology and Syntax), 357–419. To appear. Berlin: Language Science Press.

Goldberg, Adele E. 1995. Constructions: A Construction Grammar approach to ar-gument structure (Cognitive Theory of Language and Culture). Chicago, IL:The University of Chicago Press.

Goldberg, Adele E. & Ray S. Jackendoff. 2004. The English resultative as a familyof Constructions. Language 80(3). 532–568.

Greenberg, JosephH. 1963. Some universals of grammarwith particular referenceto the order of meaningful elements. In Joseph Greenberg (ed.), Universals inlanguage, 58–90. Cambridge, MA: MIT Press.

Grewendorf, Günther. 1988. Aspekte der deutschen Syntax: Eine Rektions-Bin-dungs-Analyse (Studien zur deutschen Grammatik 33). Tübingen: originallyGunter Narr Verlag now Stauffenburg Verlag.

Grewendorf, Günther. 1995. German: A grammatical sketch. In Joachim Jacobs,Arnim von Stechow, Wolfgang Sternefeld & Theo Vennemann (eds.), Syntax– Ein internationales Handbuch zeitgenössischer Forschung, vol. 2 (Handbücherzur Sprach- und Kommunikationswissenschaft 9), 1288–1319. Berlin: Walter deGruyter. DOI: 10.1515/9783110142631.2.

Grewendorf, Günther. 2002. Minimalistische Syntax (UTB für Wissenschaft: Uni-Taschenbücher 2313). Tübingen, Basel: A. Francke Verlag GmbH.

Haftka, Brigitta. 1996. Deutsch ist eine V/2-Sprache mit Verbendstellung undfreier Wortfolge. In Ewald Lang & Gisela Zifonun (eds.), Deutsch – typolo-gisch (Institut für deutsche Sprache, Jahrbuch 1995), 121–141. Berlin: Walterde Gruyter. DOI: 10.1515/9783110622522-007.

Haider, Hubert. 1986a. Fehlende Argumente: Vom Passiv zu kohärenten Infini-tiven. Linguistische Berichte 101. 3–33.

Haider, Hubert. 1986b. Nicht-sententiale Infinitive. Groninger Arbeiten zur Ger-manistischen Linguistik 28. 73–114.

Haider, Hubert. 1991. Pro-bleme? In Gisbert Fanselow & Sascha W. Felix (eds.),Strukturen und Merkmale syntaktischer Kategorien (Studien zur deutschenGrammatik 39), 121–143. Tübingen: originally Gunter Narr Verlag now Stauf-fenburg Verlag.

Haider, Hubert. 1993. Deutsche Syntax – generativ: Vorstudien zur Theorie einerprojektiven Grammatik (Tübinger Beiträge zur Linguistik 325). Tübingen: Gun-ter Narr Verlag.

Haider, Hubert. 1995. Studies on phrase structure and economy. Arbeitspapiere desSFB 340 No. 70. Stuttgart: Universität Stuttgart.


https://doi.org/10.1515/9783110142631.2

https://doi.org/10.1515/9783110622522-007

Haider, Hubert. 1997a. Projective economy: On the minimal functional structureof the German clause. InWerner Abraham & Elly van Gelderen (eds.), German:Syntactic problems—Problematic syntax (Linguistische Arbeiten 374), 83–103.Tübingen: Max Niemeyer Verlag.

Haider, Hubert. 1997b. Typological implications of a directionality constraint onprojections. In Artemis Alexiadou & T. Alan Hall (eds.), Studies on UniversalGrammar and typological variation (Linguistik Aktuell/Linguistics Today 13),17–33. Amsterdam: John Benjamins Publishing Co.

Haider, Hubert. 2020. VO/OV-base ordering. In Michael T. Putnam & B. Rich-ard Page (eds.), Cambridge handbook of Germanic linguistics (Cambridge Hand-books in Language and Linguistics 33), 339–364. Cambridge, UK: CambridgeUniversity Press. DOI: 10.1017/9781108378291.016.

Haider, Hubert & Martin Prinzhorn (eds.). 1986. Verb second phenomena in Ger-manic languages (Publications in Language Sciences 21). Dordrecht: Foris Pub-lications. DOI: 10.1515/9783110846072.

Harman, Gilbert H. 1963. Generative grammars without transformation rules: Adefense of phrase structure. Language 39(4). 597–616. DOI: 10.2307/411954.

Haugereid, Petter. 2007. Decomposed phrasal constructions. In Stefan Müller(ed.), Proceedings of the 14th International Conference on Head-Driven PhraseStructure Grammar, Stanford Department of Linguistics and CSLI’s LinGO Lab,120–129. Stanford, CA: CSLI Publications. http : / /csli - publications . stanford .edu/HPSG/2007/ (10 February, 2021).

Haugereid, Petter. 2009. Phrasal subconstructions: A Constructionalist grammardesign, exemplified with Norwegian and English (Doctoral Theses at NTNU 122).Trondheim: Norwegian University of Science & Technology. http://hdl.handle.net/11250/243990 (24 September, 2020).

Heinz, Wolfgang & Johannes Matiasek. 1994. Argument structure and case as-signment in German. In John Nerbonne, Klaus Netter & Carl Pollard (eds.),German in Head-Driven Phrase Structure Grammar (CSLI Lecture Notes 46),199–236. Stanford, CA: CSLI Publications.

Hendriksen, Hans. 1990. Sentence position of the verb in Himachali. Acta Lin-guistica Hafniensia 22(1). 159–171. DOI: 10.1080/03740463.1990.10411526.

Henriksen, Carol & Johan van der Auwera. 1994. The Germanic languages. InEkkehard König & Johan van der Auwera (eds.), The Germanic languages (Rout-ledge Language Family Descriptions 3), 1–18. London: Routledge. DOI: 10.4324/9781315812786.


https://doi.org/10.1017/9781108378291.016

https://doi.org/10.1515/9783110846072

https://doi.org/10.2307/411954

http://csli-publications.stanford.edu/HPSG/2007/

http://csli-publications.stanford.edu/HPSG/2007/

http://hdl.handle.net/11250/243990

http://hdl.handle.net/11250/243990

https://doi.org/10.1080/03740463.1990.10411526

https://doi.org/10.4324/9781315812786

https://doi.org/10.4324/9781315812786

References

Hinrichs, Erhard W & Sandra Kübler. 2005. Treebank profiling of spoken andwritten German. In Proceedings of the Fourth Workshop on Treebanks and Lin-guistic Theories (TLT). Barcelona, Spain.

Hinrichs, Erhard W. & Tsuneko Nakazawa. 1989a. Flipped out: AUX in German.In ErhardW. Hinrichs & Tsuneko Nakazawa (eds.),Aspects of German VP struc-ture (SfS-Report-01-93), 13–23. Tübingen: Universität Tübingen.

Hinrichs, Erhard W. & Tsuneko Nakazawa. 1989b. Subcategorization and VPstructure in German. In Erhard W. Hinrichs & Tsuneko Nakazawa (eds.), As-pects of German VP structure (SfS-Report-01-93), 1–12. Tübingen: UniversitätTübingen.

Hinrichs, Erhard W. & Tsuneko Nakazawa. 1994. Linearizing AUXs in Germanverbal complexes. In John Nerbonne, Klaus Netter & Carl Pollard (eds.), Ger-man in Head-Driven Phrase Structure Grammar (CSLI Lecture Notes 46), 11–38.Stanford, CA: CSLI Publications.

Hoberg, Ursula. 1981.DieWortstellung in der geschriebenen deutschen Gegenwarts-sprache (Heutiges Deutsch. Linguistische Grundlagen. Forschungen des Insti-tuts für deutsche Sprache 10). München: Max Hueber Verlag.

Höhle, Tilman N. 1978. Lexikalische Syntax: Die Aktiv-Passiv-Relation und andereInfinitkonstruktionen imDeutschen (Linguistische Arbeiten 67). Tübingen: MaxNiemeyer Verlag. DOI: 10.1515/9783111345444.

Höhle, Tilman N. 1982. Explikationen für „normale Betonung“ und „normaleWortstellung“. InWerner Abraham (ed.), Satzglieder im Deutschen – Vorschlägezur syntaktischen, semantischen und pragmatischen Fundierung (Studien zurdeutschen Grammatik 15), 75–153. Republished as Höhle (2019a). Tübingen:originally Gunter Narr Verlag now Stauffenburg Verlag.

Höhle, Tilman N. 1983. Topologische Felder. Köln, ms. Published as TopologischeFelder, 2nd edn. 2019. In Stefan Müller, Marga Reis & Frank Richter (eds.),Beiträge zur deutschen Grammatik: Gesammelte Schriften von Tilman N. Höhle(Classics in Linguistics 5), 7–89. First circulated as draft in 1983. Berlin: Lan-guage Science Press, 2019. DOI: 10.5281/zenodo.2588383.

Höhle, Tilman N. 1988. Verum-Fokus. Netzwerk Sprache und Pragmatik 5. Lund:Universität Lund, Germananistisches Institut. 2–7. Reprint as: Über Verum-Fokus im Deutschen, 2nd edn. 2019. In StefanMüller, Marga Reis & Frank Rich-ter (eds.), Beiträge zur deutschen Grammatik: Gesammelte Schriften von TilmanN. Höhle (Classics in Linguistics 5), 381–416. Berlin: Language Science Press,2019. DOI: 10.5281/zenodo.2588383.

Höhle, Tilman N. 1991. Projektionsstufen bei V-Projektionen: Bemerkungen zuF/T. Ms. Published as Höhle (2019b).


https://doi.org/10.1515/9783111345444

https://doi.org/10.5281/zenodo.2588383


Höhle, Tilman N. 1997. Vorangestellte Verben und Komplementierer sind eine na-türliche Klasse. In Christa Dürscheid, Karl Heinz Ramers & Monika Schwarz(eds.), Sprache im Fokus: Festschrift für Heinz Vater zum 65. Geburtstag, 107–120. Tübingen: Max Niemeyer Verlag. Reprint as: Vorangestellte Verben undKomplementierer sind eine natürliche Klasse, 2nd edn. 2019. In Stefan Müller,Marga Reis & Frank Richter (eds.), Beiträge zur deutschen Grammatik: Gesam-melte Schriften von Tilman N. Höhle (Classics in Linguistics 5), 417–433. Berlin:Language Science Press, 2019. DOI: 10.5281/zenodo.2588383.

Höhle, Tilman N. 2019a. Explikationen für „normale Betonung“ und „normaleWortstellung“. In Stefan Müller, Marga Reis & Frank Richter (eds.), Beiträgezur deutschen Grammatik: Gesammelte Schriften von Tilman N. Höhle, 2nd edn.(Classics in Linguistics 5), 107–191. Berlin: Language Science Press. DOI: 10 .5281/zenodo.2588383.

Höhle, Tilman N. 2019b. Projektionsstufen bei V-Projektionen: Bemerkungen zuF/T. In Stefan Müller, Marga Reis & Frank Richter (eds.), Beiträge zur deutschenGrammatik: Gesammelte Schriften von Tilman N. Höhle, 2nd edn. (Classics inLinguistics 5), 369–379. First circulated in 1991. Berlin: Language Science Press.DOI: 10.5281/zenodo.2588383.

Höhle, Tilman N. 2019c. Topologische Felder. In Stefan Müller, Marga Reis &Frank Richter (eds.), Beiträge zur deutschen Grammatik: Gesammelte Schriftenvon Tilman N. Höhle, 2nd edn. (Classics in Linguistics 5), 7–89. First circulatedas draft in 1983. Berlin: Language Science Press. DOI: 10.5281/zenodo.2588383.

Holmberg, Anders. 2015. Verb second. In Tibor Kiss & Artemis Alexiadou (eds.),Syntax – theory and analysis: An international handbook, vol. 1 (Handbooksof Linguistics and Communication Science 42), 342–383. Berlin: Mouton deGruyter. DOI: 10.1515/9783110377408.342.

Hudson, Richard. 1992. So-called ‘double objects’ and grammatical relations. Lan-guage 68(2). 251–276.

Jackendoff, Ray S. 1977. 𝑋 syntax: A study of phrase structure (Linguistic InquiryMonographs 2). Cambridge, MA: MIT Press.

Jacobson, Pauline. 1987. Review of Gerald Gazdar, Ewan Klein, Geoffrey K. Pul-lum, and Ivan A. Sag, 1985: Generalized Phrase Structure Grammar. Linguisticsand Philosophy 10(3). 389–426. DOI: 10.1007/BF00584132.

Johnson, Mark. 1986. A GPSG account of VP structure in German. Linguistics24(5). 871–882. DOI: 10.1515/ling.1986.24.5.871.

Kasper, Robert T. 1994. Adjuncts in theMittelfeld. In JohnNerbonne, Klaus Netter& Carl Pollard (eds.), German in Head-Driven Phrase Structure Grammar (CSLILecture Notes 46), 39–70. Stanford, CA: CSLI Publications.







https://doi.org/10.1515/9783110377408.342

https://doi.org/10.1007/BF00584132

https://doi.org/10.1515/ling.1986.24.5.871

References

Kim, Jong-Bok & Ivan A. Sag. 2002. Negation without head-movement. NaturalLanguage & Linguistic Theory 20(2). 339–412. DOI: 10.1023/A:1015045225019.

Kiss, Tibor. 1995. Infinite Komplementation: Neue Studien zum deutschen Verbuminfinitum (Linguistische Arbeiten 333). Tübingen: Max Niemeyer Verlag. DOI:10.1515/9783110934670.

Kiss, Tibor. 2001. Configurational and relational scope determination in German.In W. Detmar Meurers & Tibor Kiss (eds.), Constraint-based approaches to Ger-manic syntax (Studies in Constraint-Based Lexicalism 7), 141–175. Stanford, CA:CSLI Publications.

Kiss, Tibor. 2005. Semantic constraints on relative clause extraposition. NaturalLanguage & Linguistic Theory 23(2). 281–334. DOI: 10.1007/s11049-003-1838-7.

König, Ekkehard & Johan van der Auwera (eds.). 1994. The Germanic languages(Routledge Language Family Descriptions 3). London: Routledge. DOI: 10.4324/9781315812786.

Koster, Jan. 1975. Dutch as an SOV language. Linguistic Analysis 1(2). 111–136.Koster, Jan. 1999. The word orders of English and Dutch. Collective vs. individual

checking. Groninger Arbeiten zur Germanistischen Linguistik 43. 1–42.Levine, Robert D. & Walt Detmar Meurers. 2006. Head-Driven Phrase Struc-

ture Grammar: Linguistic approach, formal foundations, and computationalrealization. In Keith Brown (ed.), The encyclopedia of language and linguistics,2nd edn., 237–252. Oxford: Elsevier Science Publisher B.V. (North-Holland).DOI: 10.1016/B0-08-044854-2/02040-X.

Lohnstein, Horst. 2014. Artenvielfalt in freier Wildbahn: Generative Grammatik.In Jörg Hagemann & Sven Staffeldt (eds.), Syntaxtheorien: Analysen im Ver-gleich (Stauffenburg Einführungen 28), 165–185. Tübingen: Stauffenburg Ver-lag.

Macaulay, Monica & Colleen Brice. 1997. Don’t touch my projectile: Gender biasand stereotyping in syntactic examples. Language 73(4). 798–825.

Maling, Joan. 1990. Inversion in embedded clauses in Modern Icelandic. In JoanMaling & Annie Zaenen (eds.),Modern Icelandic syntax (Syntax and Semantics24), 71–91. Bingley, UK: Brill. DOI: 10.1163/9789004373235_004.

Meinunger, André. 2001. Restrictions on verb raising. Linguistic Inquiry 32(4).732–740.

Meurers, Walt Detmar. 1999. Raising spirits (and assigning them case). GroningerArbeiten zur Germanistischen Linguistik (GAGL) 43. 173–226. http://www.sfs.uni-tuebingen.de/~dm/papers/gagl99.html (10 February, 2021).

Meurers, Walt Detmar. 2000. Lexical generalizations in the syntax of German non-finite constructions. Arbeitspapiere des SFB 340 No. 145. Tübingen: Universi-


https://doi.org/10.1023/A:1015045225019

https://doi.org/10.1515/9783110934670

https://doi.org/10.1007/s11049-003-1838-7

https://doi.org/10.4324/9781315812786

https://doi.org/10.4324/9781315812786

https://doi.org/10.1016/B0-08-044854-2/02040-X

https://doi.org/10.1163/9789004373235_004

http://www.sfs.uni-tuebingen.de/~dm/papers/gagl99.html

http://www.sfs.uni-tuebingen.de/~dm/papers/gagl99.html

tät Tübingen. http : / /www . sfs . uni - tuebingen . de / ~dm / papers / diss . html(2 February, 2021).

Müller, Stefan. 1999. Deutsche Syntax deklarativ: Head-Driven Phrase StructureGrammar für das Deutsche (Linguistische Arbeiten 394). Tübingen: Max Nie-meyer Verlag. DOI: 10.1515/9783110915990.

Müller, Stefan. 2002. Complex predicates: Verbal complexes, resultative construc-tions, and particle verbs in German (Studies in Constraint-Based Lexicalism 13).Stanford, CA: CSLI Publications.

Müller, Stefan. 2003a. Mehrfache Vorfeldbesetzung. Deutsche Sprache 31(1). 29–62.

Müller, Stefan. 2003b. Object-to-subject-raising and lexical rule: An analysis ofthe German passive. In Stefan Müller (ed.), Proceedings of the 10th Interna-tional Conference on Head-Driven Phrase Structure Grammar, Michigan StateUniversity, 278–297. Stanford, CA: CSLI Publications. http://csli-publications.stanford.edu/HPSG/2003/mueller.pdf (10 February, 2021).

Müller, Stefan. 2006. Phrasal or lexical constructions? Language 82(4). 850–883.DOI: 10.1353/lan.2006.0213.

Müller, Stefan. 2007a. Head-Driven Phrase Structure Grammar: Eine Einführung.1st edn. (Stauffenburg Einführungen 17). Tübingen: Stauffenburg Verlag.

Müller, Stefan. 2007b. The Grammix CD Rom: A software collection for devel-oping typed feature structure grammars. In Tracy Holloway King & EmilyM. Bender (eds.), Proceedings of the Grammar Engineering Across Frameworks(GEAF07) workshop (Studies in Computational Linguistics ONLINE 2), 259–266.Stanford, CA: CSLI Publications. http://csli-publications.stanford.edu/GEAF/2007/ (2 February, 2021).

Müller, Stefan. 2010. Persian complex predicates and the limits of inheri-tance-based analyses. Journal of Linguistics 46(3). 601–655. DOI: 10 . 1017 /S0022226709990284.

Müller, Stefan. 2012. A personal note on open access in linguistics. Journal ofLanguage Modelling 0(1). 9–39. DOI: 10.15398/jlm.v0i1.52.

Müller, Stefan. 2013a. Grammatiktheorie. 2nd edn. (Stauffenburg Einführungen20). Tübingen: Stauffenburg Verlag.

Müller, Stefan. 2013b. Head-Driven Phrase Structure Grammar: Eine Einführung.3rd edn. (Stauffenburg Einführungen 17). Tübingen: Stauffenburg Verlag.

Müller, Stefan. 2013c. Unifying everything: Some remarks on Simpler Syntax,Construction Grammar, Minimalism and HPSG. Language 89(4). 920–950.DOI: 10.1353/lan.2013.0061.


http://www.sfs.uni-tuebingen.de/~dm/papers/diss.html

https://doi.org/10.1515/9783110915990

http://csli-publications.stanford.edu/HPSG/2003/mueller.pdf

http://csli-publications.stanford.edu/HPSG/2003/mueller.pdf

https://doi.org/10.1353/lan.2006.0213

http://csli-publications.stanford.edu/GEAF/2007/

http://csli-publications.stanford.edu/GEAF/2007/

https://doi.org/10.1017/S0022226709990284

https://doi.org/10.1017/S0022226709990284


https://doi.org/10.1353/lan.2013.0061

References

Müller, Stefan. 2014. Kernigkeit: Anmerkungen zur Kern-Peripherie-Unterschei-dung. In Antonio Machicao y Priemer, Andreas Nolda & Athina Sioupi (eds.),Zwischen Kern und Peripherie (studia grammatica 76), 25–39. Berlin: de Gruyter.DOI: 10.1524/9783050065335.

Müller, Stefan. 2015a. HPSG – A synopsis. In Tibor Kiss & Artemis Alexiadou(eds.), Syntax – theory and analysis: An international handbook, vol. 2 (Hand-books of Linguistics and Communication Science 42), 937–973. Berlin: Moutonde Gruyter. DOI: 10.1515/9783110363708-004.

Müller, Stefan. 2015b. Satztypen: Lexikalisch oder/und phrasal. In Rita Finkbei-ner & Jörg Meibauer (eds.), Satztypen und Konstruktionen im Deutschen (Lin-guistik – Impulse & Tendenzen 65), 72–105. Berlin: de Gruyter. DOI: 10.1515/9783110423112-004.

Müller, Stefan. 2015c. The CoreGram project: Theoretical linguistics, theory de-velopment and verification. Journal of Language Modelling 3(1). 21–86. DOI:10.15398/jlm.v3i1.91.

Müller, Stefan. 2016.Grammatical theory: From Transformational Grammar to con-straint-based approaches. 1st edn. (Textbooks in Language Sciences 1). Berlin:Language Science Press. DOI: 10.17169/langsci.b25.167.

Müller, Stefan. 2017. Head-Driven Phrase Structure Grammar, Sign-Based Con-struction Grammar, and Fluid Construction Grammar: Commonalities and dif-ferences. Constructions and Frames 9(1). 139–173. DOI: 10.1075/cf.9.1.05mul.

Müller, Stefan. 2018. A lexicalist account of argument structure: Template-basedphrasal LFG approaches and a lexical HPSG alternative (Conceptual Founda-tions of Language Science 2). Berlin: Language Science Press. DOI: 10 .5281/zenodo.1441351.

Müller, Stefan. 2019. Complex predicates: Structure, potential structure and un-derspecification. Linguistic Issues in Language Technology 16(3). 1–10.

Müller, Stefan. 2020. Grammatical theory: From Transformational Grammar toconstraint-based approaches. 4th edn. (Textbooks in Language Sciences 1). Ber-lin: Language Science Press. DOI: 10.5281/zenodo.3992307.

Müller, Stefan. 2021a. Constituent order. In Stefan Müller, Anne Abeillé, RobertD. Borsley & Jean-Pierre Koenig (eds.),Head-Driven Phrase Structure Grammar:The handbook (Empirically Oriented Theoretical Morphology and Syntax). Toappear. Berlin: Language Science Press.

Müller, Stefan. 2021b. German clause structure: An analysis with special considera-tion of so-called multiple fronting (Empirically Oriented Theoretical Morphol-ogy and Syntax). Revise and resubmit. Berlin: Language Science Press.


https://doi.org/10.1524/9783050065335

https://doi.org/10.1515/9783110363708-004

https://doi.org/10.1515/9783110423112-004

https://doi.org/10.1515/9783110423112-004


https://doi.org/10.17169/langsci.b25.167

https://doi.org/10.1075/cf.9.1.05mul




Müller, Stefan. 2021c. HPSG and Construction Grammar. In Stefan Müller, AnneAbeillé, Robert D. Borsley & Jean-Pierre Koenig (eds.), Head-Driven PhraseStructure Grammar: The handbook (Empirically Oriented Theoretical Morphol-ogy and Syntax). To appear. Berlin: Language Science Press.

Müller, Stefan, Anne Abeillé, Robert D. Borsley & Jean-Pierre Koenig (eds.). 2021.Head-Driven Phrase Structure Grammar: The handbook (Empirically OrientedTheoretical Morphology and Syntax). To appear. Berlin: Language SciencePress.

Müller, Stefan & Antonio Machicao y Priemer. 2019. Head-Driven Phrase Struc-ture Grammar. In András Kertész, Edith Moravcsik & Csilla Rákosi (eds.),Current approaches to syntax: A comparative handbook (Comparative Hand-books of Linguistics 3), 317–359. Berlin: Mouton de Gruyter. DOI: 10 . 1515 /9783110540253-012.

Müller, Stefan & Bjarne Ørsnes. 2011. Positional expletives in Danish, German,and Yiddish. In Stefan Müller (ed.), Proceedings of the 18th International Con-ference on Head-Driven Phrase Structure Grammar, University of Washington,167–187. Stanford, CA: CSLI Publications. http : / / csli - publications . stanford .edu/HPSG/2011/mueller-oersnes.pdf (10 February, 2021).

Müller, Stefan & Bjarne Ørsnes. 2013a. Passive in Danish, English, and German.In Stefan Müller (ed.), Proceedings of the 20th International Conference on Head-Driven Phrase Structure Grammar, Freie Universität Berlin, 140–160. Stanford,CA: CSLI Publications. http : / / csli - publications . stanford . edu /HPSG/2013 /mueller-oersnes.pdf (10 February, 2021).

Müller, Stefan & Bjarne Ørsnes. 2013b. Towards an HPSG analysis of object shiftin Danish. In Glyn Morrill & Mark-Jan Nederhof (eds.), Formal Grammar: 17thand 18th International Conferences, FG 2012, Opole, Poland, August 2012, revisedselected papers, FG 2013, Düsseldorf, Germany, August 2013: Proceedings (LectureNotes in Computer Science 8036), 69–89. Berlin: Springer Verlag. DOI: 10.1007/978-3-642-39998-5_5.

Müller, Stefan & StephenWechsler. 2014a. Lexical approaches to argument struc-ture. Theoretical Linguistics 40(1–2). 1–76. DOI: 10.1515/tl-2014-0001.

Müller, Stefan & Stephen Wechsler. 2014b. Two sides of the same slim Boojum:Further arguments for a lexical approach to argument structure. TheoreticalLinguistics 40(1–2). 187–224. DOI: 10.1515/tl-2014-0009.

Muysken, Pieter. 1982. Parametrizing the notion of “head”. Journal of LinguisticResearch 2. 57–75.

Nerbonne, John. 1986. ‘Phantoms’ and German fronting: Poltergeist con-stituents? Linguistics 24(5). 857–870. DOI: 10.1515/ling.1986.24.5.857.


https://doi.org/10.1515/9783110540253-012

https://doi.org/10.1515/9783110540253-012

http://csli-publications.stanford.edu/HPSG/2011/mueller-oersnes.pdf




https://doi.org/10.1007/978-3-642-39998-5_5

https://doi.org/10.1007/978-3-642-39998-5_5

https://doi.org/10.1515/tl-2014-0001

https://doi.org/10.1515/tl-2014-0009

https://doi.org/10.1515/ling.1986.24.5.857

References

Netter, Klaus. 1992. On Non-Head Non-Movement: An HPSG Treatment of Fini-te Verb Position in German. In Günther Görz (ed.), Konvens 92. 1. Konferenz„Verarbeitung natürlicher Sprache“. Nürnberg 7.–9. Oktober 1992 (Informatik ak-tuell), 218–227. Berlin: Springer Verlag. DOI: 10.1007/978-3-642-77809-4.

Netter, Klaus. 1998. Functional categories in an HPSG for German (SaarbrückenDissertations in Computational Linguistics and Language Technology 3). Saar-brücken: Deutsches Forschungszentrum für Künstliche Intelligenz and Univer-sität des Saarlandes.

Oppenrieder, Wilhelm. 1991. Von Subjekten, Sätzen und Subjektsätzen (Linguisitis-che Arbeiten 241). Tübingen: Max Niemeyer Verlag.

Ørsnes, Bjarne. 2009. Das Verbalfeldmodell: Ein Stellungsfeldermodell für denkontrastiven DaF-Unterricht. Deutsch als Fremdsprache 46(3). 143–149.

Pabst, Katharina, Paola Cépeda, Hadas Kotek, Kristen Syrett, Katharine Donelson& Miranda McCarvel. 2017. Gender bias in linguistic textbooks: Has anythingchanged since macauly & brice (1997). Slides of a talk given at the LSA annualmeeting 2017. https://paolacepeda.files.wordpress.com/2018/01/lsa2018_pabst-et-al-_gender-bias-in-linguistics-textbooks.pdf (15 June, 2020).

Penn, Gerald. 2004. Balancing clarity and efficiency in typed feature logicthrough delaying. In Donia Scott (ed.), Proceedings of the 42nd Meeting of theAssociation for Computational Linguistics (ACL’04), 239–246. Barcelona, Spain:Association for Computational Linguistics. DOI: 10.3115/1218955.1218986.

Perlmutter, DavidM. 1978. Impersonal passives and the Unaccusative Hypothesis.In Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society, 157–189. Berkeley Linguistic Society. https://escholarship.org/uc/item/73h0s91v(26 April, 2018).

Plank, Frans (ed.). 2003. Das grammatische Raritätenkabinett. Universität Kon-stanz. https://typo.uni-konstanz.de/rara/intro/ (27 April, 2020).

Poletto, Cecilia. 2002. The left-periphery of V2-Rhaetoromance dialects: A newview on V2 and V3. In Sjef Barbiers, Leonie Cornips & Susanne van der Kleij(eds.), Syntactic microvariation, 214–242. Meertens Instituut. https : / / www .meertens.knaw.nl/books/synmic/pdf/poletto.pdf (27 April, 2020).

Pollard, Carl J. 1996. On head non-movement. In Harry Bunt & Arthur van Horck(eds.), Discontinuous constituency (Natural Language Processing 6), 279–305.Veröffentlichte Version eines Ms. von 1990. Berlin: Mouton de Gruyter. DOI:10.1515/9783110873467.279.

Pollard, Carl & Ivan A. Sag. 1987. Information-based syntax and semantics (CSLILecture Notes 13). Stanford, CA: CSLI Publications.


https://doi.org/10.1007/978-3-642-77809-4

https://paolacepeda.files.wordpress.com/2018/01/lsa2018_pabst-et-al-_gender-bias-in-linguistics-textbooks.pdf

https://paolacepeda.files.wordpress.com/2018/01/lsa2018_pabst-et-al-_gender-bias-in-linguistics-textbooks.pdf

https://doi.org/10.3115/1218955.1218986

https://escholarship.org/uc/item/73h0s91v

https://typo.uni-konstanz.de/rara/intro/

https://www.meertens.knaw.nl/books/synmic/pdf/poletto.pdf

https://www.meertens.knaw.nl/books/synmic/pdf/poletto.pdf

https://doi.org/10.1515/9783110873467.279

Pollard, Carl & Ivan A. Sag. 1994.Head-Driven Phrase Structure Grammar (Studiesin Contemporary Linguistics 4). Chicago, IL: The University of Chicago Press.

Prince, Ellen F. 1989. Yiddish Wh-clauses, subject-postposing, and topicalization.In Joyce Powers & Kenneth de Jong (eds.), Proceedings of ESCOL 5 1988, 403–415. Ohio State University.

Przepiórkowski, Adam. 1999. On case assignment and “adjuncts as complements”.In Gert Webelhuth, Jean-Pierre Koenig & Andreas Kathol (eds.), Lexical andconstructional aspects of linguistic explanation (Studies in Constraint-BasedLexicalism 1), 231–245. Stanford, CA: CSLI Publications.

Przepiórkowski, Adam & Anna Kupść. 2006. HPSG for Slavicists. Glossos 8. 1–68.Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A

comprehensive grammar of the English language. London: Longman.Reis, Marga. 1980. On justifying topological frames: ‘Positional field’ and the or-

der of nonverbal constituents in German. Documentation et Recherche en Lin-guistique Allemande Contemporaine. Revue de Linguistique 22/23. 59–85. DOI:10.3406/drlav.1980.957.

Rizzi, Luigi. 1990. Speculations on verb-second. In JoanMascaró&MarinaNespor(eds.), Grammar in progress: GLOW essays for Henk van Riemsdijk (Studies inGenerative Grammar 36), 375–386. Dordrecht: Foris Publications. DOI: 10.1515/9783110867848.375.

Roberts, Ian. 1993. Verbs and diachronic syntax: A comparative history of Englishand French (Studies in Natural Language and Linguistic Theory 28). Dordrecht:Kluwer Academic Publishers. DOI: 10.1007/978-94-011-2910-7.

Robinson, Orrin W. 1992. Old English and its closest relatives: A survey of the ear-liest Germanic languages. Stanford, CA: Stanford University Press.

Ross, MalcolmD. 2004. Themorphosyntactic typology of oceanic languages. Lan-guage and Linguistics 5(2). 491–541.

Sag, Ivan A. 1997. English relative clause constructions. Journal of Linguistics33(2). 431–483. DOI: 10.1017/S002222679700652X.

Sag, Ivan A. 2012. Sign-Based Construction Grammar: An informal synopsis. InHans C. Boas & Ivan A. Sag (eds.), Sign-Based Construction Grammar (CSLILecture Notes 193), 69–202. Stanford, CA: CSLI Publications.

Sag, Ivan A., Hans C. Boas & Paul Kay. 2012. Introducing Sign-Based Construc-tion Grammar. In Hans C. Boas & Ivan A. Sag (eds.), Sign-Based ConstructionGrammar (CSLI Lecture Notes 193), 1–29. Stanford, CA: CSLI Publications.

Sag, IvanA., Rui Chaves, AnneAbeillé, Bruno Estigarribia, FrankVan Eynde, DanFlickinger, Paul Kay, Laura A. Michaelis-Cummings, Stefan Müller, Geoffrey K.


https://doi.org/10.3406/drlav.1980.957

https://doi.org/10.1515/9783110867848.375

https://doi.org/10.1515/9783110867848.375

https://doi.org/10.1007/978-94-011-2910-7

https://doi.org/10.1017/S002222679700652X

References

Pullum & Thomas Wasow. 2020. Lessons from the English auxiliary system.Journal of Linguistics 56(1). 87–155. DOI: 10.1017/S002222671800052X.

Sag, Ivan A., Thomas Wasow & Emily M. Bender. 2003. Syntactic theory: A for-mal introduction. 2nd edn. (CSLI Lecture Notes 152). Stanford, CA: CSLI Publi-cations.

Scherpenisse, Wim. 1986. The connection between base structure and linearizationrestrictions in German and Dutch (Europäische Hochschulschriften, Reihe XXI,Linguistik 47). Frankfurt/M.: Peter Lang.

Stahlberg, Dagmar, Sabine Sczesny & Friederike Braun. 2001. Name your fa-vorite musician: Effects of masculine generics and of their alternatives in Ger-man. Journal of Language and Social Psychology 20(4). 464–469. DOI: 10.1177/0261927X01020004004.

Storto, Luciana R. 2003. Interactions between verb movement and agreement inKaritiana (Tupi stock). Revista Letras 60. 411–433.

Tesnière, Lucien. 2015. Elements of structural syntax. Translated by Timothy Os-borne and Sylvain Kahane. Amsterdam: John Benjamins Publishing Co. DOI:10.1075/z.185.

Thráinsson, Höskuldur. 2007. The syntax of Icelandic (Cambridge Syntax Guides).Cambridge, UK: Cambridge University Press. DOI: 10.1017/CBO9780511619441.

Uszkoreit, Hans. 1987. Word order and constituent structure in German (CSLI Lec-ture Notes 8). Stanford, CA: CSLI Publications.

Vance, Barbara S. 1997. Syntactic change in Medieval French: Verb-second and nullsubjects (Studies in Natural Language and Linguistic Theory 41). Dordrecht:Kluwer Academic Publishers. DOI: 10.1007/978-94-015-8843-0.

Wechsler, Stephen. 1995. The semantic basis of argument structure (Dissertationsin Linguistics). Stanford, CA: CSLI Publications.

Weiß, Helmut. 2005. Inflected complementizers in Continental West Germanicdialects. Zeitschrift für Dialektologie und Linguistik 72(2). 148–166.

Weiß, Helmut. 2018. The Wackernagel Position and complementizer agreement:The emergence of a syntactic particularity at the left edge of the German mid-dle field. In Agnes Jäger, Gisella Ferraresi & Helmut Weiß (eds.), Clause struc-ture and word order in the history of German (Oxford Studies in Diachronic andHistorical Linguistics), 132–154. Oxford: Oxford University Press.

Willis, David W. E. 1998. Syntactic change in Welsh: A study of the loss of verb-second. Oxford: Clarendon Press.

Wurmbrand, Susanne. 2003a. Infinitives: Restructuring and clause structure (Stud-ies in Generative Grammar 55). Berlin: Mouton de Gruyter. DOI: 10 . 1515 /9783110908329.


https://doi.org/10.1017/S002222671800052X

https://doi.org/10.1177/0261927X01020004004

https://doi.org/10.1177/0261927X01020004004

https://doi.org/10.1075/z.185

https://doi.org/10.1017/CBO9780511619441

https://doi.org/10.1007/978-94-015-8843-0

https://doi.org/10.1515/9783110908329

https://doi.org/10.1515/9783110908329

Wurmbrand, Susanne. 2003b. Long passive (corpus search results). Ms. Universityof Connecticut.

Yip, Moira, Joan Maling & Ray Jackendoff. 1987. Case in tiers. Language 63(2).217–250. DOI: 10.2307/415655.

Zaenen, Annie, Joan Maling & Höskuldur Thráinsson. 1985. Case and grammati-cal functions: The Icelandic passive.Natural Language & Linguistic Theory 3(4).441–483. DOI: 10.1007/BF00133285.


https://doi.org/10.2307/415655

https://doi.org/10.1007/BF00133285

Name index

Ackerman, Farrell, 140Adams, Marianne, 20Allwood, Jens, viiAlsina, Alex, 66Anderson, Stephen R., 20Askedal, John Ole, 167Asudeh, Ash, 66

Bartsch, Renate, 112Bech, Gunnar, 95, 167Bender, Emily M., 70Berman, Judith, 56Bhatt, Rakesh Mohan, 20Bierwisch, Manfred, 109Bildhauer, Felix, 191Blackburn, Patrick, 61Bloomfield, Leonard, viBoas, Hans C., 66Bonami, Olivier, 113Borsley, Robert D., 20, 70, 182Bos, Johan, 61Bouma, Gosse, 22Bresnan, Joan, 55Brice, Colleen, vii, viii

Chomsky, Noam, vi, 11, 33, 55, 56Christie, Elizabeth, 66Copestake, Ann, vCrysmann, Berthold, 112, 182Culicover, Peter W., 66

Davis, Anthony R., 88De Kuthy, Kordula, vden Besten, Hans, 9Diesing, Molly, 25–27, 174, 176, 180Dowty, David, 88Dryer, Matthew S., 12–14

Duden, 188

Eisenberg, Peter, 43, 52, 72, 188

Fanselow, Gisbert, 90, 102, 122Fillmore, Charles J., 47Fitch, W. Tecumseh, 2Flickinger, Dan, vFontana, Josep M., 20Frey, Werner, 89, 122Fries, Norbert, 126

Gazdar, Gerald, 55, 66, 76, 103Geach, Peter Thomas, 96Ginzburg, Jonathan, 183, 191Godard, Danièle, 105Goldberg, Adele E., 66Greenberg, Joseph H., 11, 13Grewendorf, Günther, 89, 155

Haftka, Brigitta, 11Haider, Hubert, 16, 20, 56, 72, 110, 145,

153, 161, 163, 167, 168, 185Harman, Gilbert H., 61Haugereid, Petter, 66Heinz, Wolfgang, 153Hendriksen, Hans, 20Henriksen, Carol, 1, 7Hinrichs, Erhard W, 108Hinrichs, Erhard W., 19, 96, 105, 160Hoberg, Ursula, 18Höhle, Tilman N., 13, 15, 21, 67, 110, 119,

163, 224Holmberg, Anders, 20Hudson, Richard, 18

Jackendoff, Ray S., vi, 30, 33, 55, 66, 148

Name index

Jacobson, Pauline, 66Johnson, Mark, 66

Kasper, Robert T., 112Kathol, Andreas, 20Kay, Paul, 66Kiss, Tibor, 61, 72, 90, 91, 97, 102, 105Klein, Ewan, 66Koenig, Jean-Pierre, 88König, Ekkehard, 1Koster, Jan, 22, 112Kübler, Sandra, 108Kupść, Anna, 191

Levine, Robert D., 191Lohnstein, Horst, 89

Macaulay, Monica, vii, viiiMachicao y Priemer, Antonio, 191Maling, Joan, 25, 30, 132–134, 141, 148,

174, 185Matiasek, Johannes, 153Meinunger, André, 110Meurers, Walt Detmar, 105, 148, 191Müller, Stefan, v, vii, 11, 27, 33, 34, 36,

60, 61, 66, 73, 74, 76, 88, 91, 95,105, 109, 127, 153, 159, 163, 177,181, 182, 184, 189, 191, 207

Muysken, Pieter, 57

Nakazawa, Tsuneko, 19, 96, 105, 160Nerbonne, John, 66Netter, Klaus, 61, 111

Oppenrieder, Wilhelm, 163Ørsnes, Bjarne, 15, 27, 74, 113, 159, 177,

184, 189

Pabst, Katharina, viiiPenn, Gerald, vPerlmutter, David M., 153Plank, Frans, 20Poletto, Cecilia, 20Pollard, Carl, v, ix, 70, 72, 88, 182, 191Prince, Ellen F., 28, 178

Prinzhorn, Martin, 20Przepiórkowski, Adam, 148, 191Pullum, Geoffrey K., 66

Quirk, Randolph, 23

Reis, Marga, 112Rizzi, Luigi, 20Roberts, Ian, 20Robinson, Orrin W., 7Ross, Malcolm D., 20

Sag, Ivan A., v, ix, 66, 70, 88, 103, 182,183, 191

Samvelian, Pollet, 105Scherpenisse, Wim, 117Stahlberg, Dagmar, ixStorto, Luciana R., 20

Tallerman, Maggie, 20Tesnière, Lucien, 63, 64Thráinsson, Höskuldur, 132–134, 138,

141, 185

Uszkoreit, Hans, 55, 66

van der Auwera, Johan, 1, 7Vance, Barbara S., 20Vennemann, Theo, 112

Wasow, Thomas, 70Wechsler, Stephen, 66, 88, 158Weiß, Helmut, 119Willis, David, 20Willis, David W. E., 20Wurmbrand, Susanne, 102, 164

Yip, Moira, 30, 148

Zaenen, Annie, vi, x, 30, 132–134, 141,185


Language index

Afrikaans, 9Aramaic, 9

Breton, 20Burgundian, 5

Cornish, 20

Danish, 6, 19, 27, 28, 30, 744, 81, 128, 129,137–140, 171, 177, 178, 198, 208,213

Dutch, 9, 22, 81

English, 10, 30, 479, 54, 78, 171Estonian, 20

Faroese, 7Frisian, 10

German, 8, 18–20, 22, 27, 29–31, 81, 83,87, 92, 93, 104, 126, 1265, 127–129, 131–136, 138, 139, 142, 143,160, 163, 165, 171, 173, 175–177,184, 188, 197, 204, 208, 210, 213

Gothic, 5

Hebrew, 9Himachali, 20

Icelandic, 7, 25, 132–135, 138, 171, 174, 185Inuit, 6

Karitiana, 20

Norwegian, 7

Old French, 20Old Spanish, 20

Pennsylvania German, 9

Rhaeto-Romance, 20

Sisiqa, 20Sorbian, 20Swedisch, 6

Taiof, 20

Welsh, 20

Yiddish, 9, 25, 28, 174, 178, 180, 187

Draft of 20th October 2021, 18:56

Germanic syntax

This book is an introduction to the syntactic structures that can be found in the Germanic lan-guages. The analyses are couched in the framework of HPSG light, which is a simplified versionof HPSG that uses trees to depict analyses rather than complicated attribute value matrices.

The book is written for students with basic knowledge about case, constituent tests, and sim-ple phrase structure grammars (advanced BA or MA level) and for researchers with an interestin the Germanic languages and/or an interest in Head-Driven Phrase Structure Grammar/Sign-Based Construction Grammar without having the time to deal with all the details of these theo-ries.

Stefan Müller - hu-berlin.de

Documents

Transcript of Stefan Müller - hu-berlin.de